JPH03172950A - Management host switching control method - Google Patents

Abstract

PURPOSE:To quickly continue communication by automatically switching a management host at the time of fault of the management host. CONSTITUTION:When receiving the declaration of the management host, a communication control processing device 11 sets this request to a stack 12 and reports permission to the host processor which first declares, for example, a host processor 10A, and hereafter, the host processor 10A is operated as the management host. If trouble occurs in the host processor 10A as the management host, the communication control processing device 11 selects a host processor 10B, which declares the management host just after the host processor 10A, from the stack 12 and reports permission to the host processor 10B. Thus, the management host is automatically switched at the time of fault of the management host.

Description

[Detailed Description of the Invention] (Summary) Management in a system in which a plurality of host processing devices share a communication control processing device, and a management host that manages communication resources under the communication control processing device is defined. Regarding host switching control method.

The purpose is to allow other host processing devices to quickly continue communication in the event that a failure occurs in the management host.

The process of notifying the declaration of the management host from each host processing device that has the function of becoming a management host to the communication control processing device, and the process of stacking the requests when one communication control processing device receives the declaration of the management host. , the process of notifying the management host permission to the host processing device that declared earlier, and the process of selecting another host processing device from the stack that has declared the management host next if the management host fails. , and a process of newly notifying management host permission, and is configured to automatically switch the management host in the event of a failure of the management host.

[Industrial application field]

The present invention relates to a management host switching control method in a system in which a plurality of host processing devices share a communication control processing device, and in which a management host that manages communication resources under the communication control processing device is determined.

[Conventional technology]

FIG. 3 shows an example of a conventional technique, and FIG. 4 shows an example of a conventional command sequence.

As shown in FIG. 3(a), a plurality of host processing devices 10
A, IOB is one communication control processing unit (CCP) 11
When the same communication line 13 under the communication control processing device 11 is shared, the control is conventionally performed as follows, for example.

■ One host predetermined as a management host, for example, the host processing device 10A, activates one communication line 13 or terminal 20.
The host processing device 10A is stored as a management host.

■ When a session establishment request is issued from the terminal 20.

The communication control processing device 11 notifies the host processing device 10A that manages the terminal 20 of a session establishment request message.

■ The host processing device 10A is the management host.

The terminal 20 determines the communication party requested by looking at the message, and if it is a program running on the host processing device 10B, the session establishment request is sent to the host processing device 10B.
to notify.

(2) Upon receiving the notification, the host processing device 10B establishes a logical connection and performs data communication with the terminal 20.

As shown in FIG. 3(b), if the host processing device 10A, which is the management host, goes down due to a failure, it will no longer be able to accept requests from the terminal 20.

Therefore, in the conventional technology, when the host processing device 10A becomes in trouble, an operator intervenes and the host processing device 10B
From now on, the communication control processing device 11 causes the terminal management notification shown as ■ to be performed.

It memorizes that the management host of the terminal 20 is the host processing device 10B, and notifies the session establishment request from the terminal 20 to the host processing device 10B.

FIG. 4 shows the command sequence between these host processing devices and communication control processing devices. Similar to the example in Figure 3, the management host is host A, and the remote terminal is host)
It is assumed that communication with B is made.

■ When host A, which is the management host, is started up, host A notifies the other terminal that the administrator of the terminal is host A using the ACTPU/ACTLυ command. This is a command to activate a physical unit/logical unit that is a communication resource.

+resp is the response to it.

(2) When a communication request for host 1-B is issued to the other terminal, the other terminal sends a session establishment request message to host A because the management host is host A.

■ Host A determines that the session establishment request is for host B, and issues the CDINIT command.
Host B is notified that a connection establishment request has been made.

The CDINIT command is a communication command between hosts.

■ Host B sends a session start message (BIND command) to the other party's terminal. Thereafter, data communication is performed between host B and the other party's terminal.

■ If a failure occurs on host A, which is the management host,
In order to switch the management host from host A to host B according to instructions from the operator,
Issue the TPLI/ACTLU command.

[Problem to be solved by the invention]

Conventionally, when multiple host processing devices share a communication control processing device and use communication resources such as communication lines under the communication control processing device, the management host is determined in advance, and the communication is performed via the management host. Establish a logical session with one communication partner.

It is designed to communicate.

Therefore, there is a problem in that when a failure occurs in the management host, other host processing devices cannot establish a new logical session. Also.

In order to make this possible, an operation is required to instruct another host processing device to become a new management host, which poses a problem in that communication cannot be restarted quickly.

An object of the present invention is to solve the above-mentioned problems, and to enable communication to be quickly continued using another host processing device when a failure occurs in the management host.

[Means to solve the problem]

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

In FIG. 1, IOA and IOB are host processing devices equipped with a CPU and memory and perform data processing, and 11 is a communication control processing device (CCP) that controls communication via a communication line.
), 12 represents a stack that stores management host declaration information.

Host processing device 10A, IOB is 1 communication control processing device 1
1 is shared.

When starting up the system of the host processing device 10A.

In process P1, the host processing device 10A receives the cell establishment request message from the terminal and declares a management host that notifies the two communication partners of the request. At startup, processing P
Step 2 declares the management host.

When the communication control processing device 11 receives the declaration from the management host, it is a host processing device that sets the request in the stack 12 and declares it early in process P3.

For example, the host processing device 10A is notified of management host permission.

Thereby, the host processing device 10A operates as a management host from now on.

When the host processing device 10A that has become the management host becomes a failure, the communication control processing device 11 selects the host processing device 10B that is declaring the management host next from the host processing device 10A from the stack 12 in process P4. , newly notifies the host processing device 10B of management host permission.

As a result, in the event of a failure of the management host, the management host can be automatically switched without the intervention of an operator or the like.

[Function] In the present invention, only one communication control processing device! Since Ell stores the declarations of all management hosts in a stacked manner and notifies permission of a new management host when a failure occurs in a management host, automatic continuation of one communication becomes possible.

Note that information regarding the host processing device granted permission as a management host may be stored separately from the stack 12 shown in FIG.

〔Example〕

FIG. 2 shows an example command sequence according to one embodiment of the invention.

In Figure 2, the sequence indicated by the thick solid line is the fourth
This is a different part from the conventional technology explained in the figure. The following description assumes that two host processing devices, host A and host B, share a communication control processing device, and a partner terminal communicates with host B.

■ When starting the system, host A declares that it can become a management host before issuing the ACTPU/ACTLU command to notify that it owns the other terminal.
This command is called the ACTDTE command.

■ The communication control processing unit returns a response of 10 resp, stacks the declaration of this management host, and, since this declaration is the first declaration, grants permission for the management host to host A.
give to

■ After obtaining permission from the management host, host A sends a notification to the other terminal that host A is the administrator of the terminal.
Performed by CTPU/ACTLU=+ManF.

(2) When a communication request to host B is issued to the other terminal, the other terminal sends a session establishment request message to host A because the managing host is boss A.

■ Host A determines that the session establishment request is for host B, and issues the CDrNIT command.
Notify host B that a session establishment request has been made.

■ Host) B sends a session start message (BIND command) to the other party's terminal. From then on.

Data communication is performed between host) B and the other party's terminal.

■ Although the explanation is different in terms of time, host B, which started up later than host A, is the same as host A.

The 0 communication control processing device that issues the ACTDTE command and declares the management host stacks this declaration.

However, since the management host permission has already been given to host A, this permission is not given to host B.

Suppose that a failure occurs in host A after that. A host failure may be detected directly by the two communication control processing units, or may be detected by a notification from the host processing unit or a monitoring device. Various methods have been known for this failure detection. So.

The explanation here will be omitted.

■ When the failure of host A is recognized, the communication control processing unit 1 retrieves the stack information and sends it to host B.
Since it can be seen that host B has declared the management host, host B is granted permission to be the management host. This is done automatically without operator intervention. Upon receiving this permission, host B notifies host B that the owner of the other party's terminal is host B.

Be notified.

Although an example has been described in which two host processing devices share a communication control processing device, the same control is possible with three or more host processing devices.

〔Effect of the invention〕

As explained above, (1) according to the present invention, in the event of a host failure, it is possible to automatically switch to an alternative host without performing an operation for switching the management host, and (2) it is possible to prevent communication from being stopped. .

[Brief explanation of the drawing]

FIG. 1 is a diagram explaining the principle of the present invention. FIG. 2 is an example of a command sequence according to an embodiment of the present invention. Figure 3 is an example of conventional technology. FIG. 4 shows an example of a conventional command sequence. In the figure, IOA and IOB are host processing units, 11 is a communication control processing unit, 12 is a stack, 13 is a communication line, and PI-
P4 represents a processing process.

Claims (1)

[Claims] In a system in which a plurality of host processing devices (10A, 10B) share a communication control processing device (11), a message from a terminal requesting logical connection with a communication partner is received and the request is made. The process of notifying the communication control processing device from each host processing device that has the ability to become a management host of the declaration of the management host to be distributed (P1, P2), and when the communication control processing device receives the declaration of the management host. The process of stacking the request and notifying the management host permission to the host processing device that declared earlier (P3
), and if the host processing unit that has become the management host fails, the communication control processing unit selects another host processing unit from the stack that is declaring itself as the management host next, and assigns the command to that host processing unit. A management host switching control method comprising the step of notifying a new management host permission (P4) to automatically switch the management host in the event of a failure of the management host.