JPH0275046A - Control system for node active state - Google Patents

Abstract

PURPOSE:To avoid the transmission of the useless commands to the inactive nodes by enabling each host to monitor by itself the active states of the nodes connected to other hosts without having any inquiry to other hosts. CONSTITUTION:The hosts A and B contain the mechanisms 2 and 5 to always monitor the active/inactive states of the nodes (a) - (f) connected to both hosts and also to other hosts. Therefore the hosts A and B have the state control information on the nodes (a) - (f) which are not directly connected to both hosts in a network system. Thus it is possible to know whether those nodes (a) - (f) are active or not in the network by reference to the state control information even in the case a command is sent to any one of nodes (a) - (f). As a result, the optimum process is possible in accordance with the active/inactive states of the nodes (a) - (f) that received the commands.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a method for managing the active state of nodes, and in particular, when a plurality of host computer systems (hereinafter referred to as hosts) are interconnected and each host The present invention relates to a method for managing the active state of nodes in a distributed network system having a multiple host connection configuration in which nodes such as multiple distributed processor systems and packet switching systems are connected.

[Conventional technology]

-IIIQ, in a distributed network system consisting of multiple hosts and multiple nodes connected to these hosts, each host can only access the nodes that are directly connected to its own host. Managing the state. Therefore, in such a conventional bouncing system, when a host issues a command request to a node connected to another host, that host can activate/deactivate the node making the command request. Since the status cannot be known, commands are sent regardless of whether the node is active or inactive.

For this reason, in the prior art, if the node of the command light from the host is inactive, the post will send a useless command, and the response may vary depending on the operating status of the other host that received the command. However, there are cases where the service does not return immediately, resulting in a slow response to the requester.

In addition, as conventional technologies related to such distributed networks, for example, "Application of rO3I - Toward Realization of Advanced Information Technology" (Japanese Standards Association, Akio Munekami et al., published in 1987, p. 24), "Information Theory of Work (Iwahi Lecture Information Science-5, Iwanami Shoten, Masa Noro et al., 19
The technique described in Kasa (published in 1982, pages 178 and 251 to 255) is known.

[Problem to be solved by the invention]

The above-mentioned conventional technology takes into consideration the fact that a certain host controls the transmission of commands to nodes connected to other hosts by managing the activation status of nodes connected to other hosts. It has not been. Therefore, in the conventional technology, when a certain host needs to send a command to a node connected to another host,
Even if this node is in an inactive state, the command will be sent, resulting in useless command transmission and reporting that the command cannot be executed because the command receiving node is in an inactive state. There is a problem in that there are cases where the data cannot be immediately returned to the requester.

It is an object of the present invention to solve the problems of the prior art described above, to suppress the transmission of unnecessary commands, and, when it is impossible to execute a command, to immediately return a report of the impossibility to the request source. An object of the present invention is to provide a method for managing the active state of nodes in a distributed network system.

[Means to solve the problem]

According to the present invention, the above object is achieved by enabling each host to monitor the active status of nodes connected to other hosts without inquiring other hosts.

For this reason, each host is equipped with state management information for each node that indicates the active/inactive state of nodes connected to each host, including other hosts.
During negotiation processing before the start of communication, the state management information provided in each host is exchanged.

Also, if the status of a node changes after communication starts,
A host that is directly connected to that node detects changes in the node's state, updates state management information in its own host, and creates a protocol to report this node's state changes to other hosts. I try to do that.

Other hosts that receive protocol data reporting this state change can update the state management information managed by the host itself.

[Operation] Each host in the network system has state management information even for nodes that are not directly connected to the host, so it is possible to issue commands to any node within the network. Even when sending a command, you can know whether the node is active or not by referring to the state management information, and if the node to which the command is sent is inactive, the command will be sent immediately. , it becomes possible to perform optimal processing according to the active/inactive state of the node to which the command is sent, such as reporting to the request source of the command that the command cannot be sent.

As a result, according to the present invention, it is possible to prevent the host from sending unnecessary commands, and it is possible to promptly report to the command requester that the command cannot be transmitted. .

〔Example〕

Hereinafter, an embodiment of the node activation state management method according to the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a multi-host connection distributed network system to which the present invention is applied. In FIG. 1, A and B are host computer systems (host I-), a to f are nodes, 1.4 is a node status change detection mechanism, 2.5 is a node status management mechanism, and 3.6
is an information transfer mechanism, and 7 is a communication line.

For example, as shown in FIG. 1, a distributed network system with a multiple host connection configuration to which the present invention is applied includes two hosts A and B connected to each other by a communication line 7, and the respective hosts A and B. It is composed of a plurality of nodes a-C, d, and -f connected to each other by a communication line 7.

In the node activation state management method of the present invention applied to such a network system, each of hosts A and B can check the activation/inactivation state of a node under the control of the other host on its own host. It also makes it possible to manage. For this reason, each host A, B has a node state change detection mechanism 1, 4 that detects changes in the active/inactive state of nodes connected to the host itself, and controls node active state management information. The node status management mechanism 2.5 is configured to include a communication control mechanism 3.6 for exchanging node activation status management information and node activation/inactivation status change information in the host phase 'Ti1E.

The node state change detection mechanism 1.4 operates according to a change in the active/inactive state of a node directly connected to its host, and the node state management mechanism 2.5 and information transfer mechanism 3.6 The node state change detection mechanism 1, 4
It operates to reflect the detection of a node state change by the node in the active state management information of the node. That is, the node status management mechanisms 2 and 5 have the function of recording the active/inactive status of nodes connected to the host, and the information transfer mechanisms 3 and 6 have the function of recording the active/inactive status of the nodes connected to the host, and the information transfer mechanisms 3 and 6 have the function of recording the active/inactive status of the nodes connected to the host, and the information transfer mechanisms 3 and 6 During the preceding negotiation process, the host A sends the node configuration information of the nodes connected to each host to the other host via the communication line 7.

After B becomes communicable, if there is a change in the activation/deactivation of a node connected to each host, it has the function of transmitting the status change to the other host via the communication line 7. .

In the network system not shown in FIG. 1, it is assumed that there are two posts, but a larger number of hosts may be provided. Also, the number of nodes is arbitrary, and these nodes can be distributed processor systems or
It may be a packet switching system or the like, and is assumed to have intelligence that can operate based on commands from a host.

Hereinafter, the operation of one embodiment of the present invention configured as described above will be explained in detail.

First, hosts A and B perform negotiation processing for the purpose of checking protocol compatibility between hosts A and B in order to enable communication between hosts A and B. At that time, each host) is directly connected to A and B, that is,
Connected node configuration information of nodes a-c, d=f managed by each host A, B and activation status management information for each node are mutually exchanged via information transfer mechanisms 3, 6 and communication line 7. and registered in the node state management mechanism 2.5.

This allows hosts A and B to communicate with each other.

After that, nodes a-c connected to hosts l-A and B
, when a change occurs in the active/inactive state of dxf, the node state change detection mechanism 1 in each host A, B
．． 4, the state change is detected. The detection information indicating this state change updates the active state management information of No-Y managed by each host in the node state management mechanisms 2 and 5, and reports the state change to the other host via the communication line 7. Sent as .

When the information transfer mechanism 3.6 in each host A, B receives the node activation/deactivation status change report sent from the other host, the node status management mechanism 2.5 managed by the own host The activation state management information of the corresponding node in the node is updated.

In the embodiment of the present invention that operates as described above, the node status management mechanism 2.5 provided in each host stores a node status indicating the active/inactive status of all nodes accommodated in the network. Since the state management information is always updated and held in the latest state, each host can optimally control commands and the like to each node based on this information.

That is, according to the embodiment of the present invention described above, when a certain host l- makes an operation request to a node connected to another host, that host changes the state of the node managed by the host itself. By checking the activation status management information of the corresponding nodes in the management mechanisms 2 and 5, it is possible to know the activation/inactivation status of nodes connected to other hosts at the other end, and whether the node is in an inactive status or not. If so, it is possible to stop sending the command to the destination host and immediately report to the request source that the command cannot be executed. Further, according to an embodiment of the present invention,
Each host activates/activates nodes connected to other hosts.
It becomes possible to know the inactive state without inquiring other hosts.

[Effects of the Invention] As explained above, according to the present invention, each host can know the active/inactive status of nodes connected to other hosts without making inquiries to other hosts. , it is possible to prevent unnecessary transmission of commands to nodes in an inactive state.

Further, according to the present invention, since unnecessary transmission of commands can be prevented, it is possible to quickly respond to the command request source. Furthermore, since the own host has management information for nodes connected to other hosts, it is possible to monitor the status of nodes managed by other hosts without making inquiries to other hosts. This has the effect of making it possible.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of a multi-host connection distributed network system to which the present invention is applied. A, B--host computer system (host),
af-・-1---Node, 1.4---Node status change detection mechanism, 2, 5---Node status management mechanism, 3.6---Information transfer mechanism, 7 −−−−−−A communication line. Figure 1 A, B + host controller J stem (/'I'I ST) 7-71-I To I plan

Claims (1)

[Claims] 1. In a distributed network system with a plurality of hosts connected, each host is equipped with a mechanism for constantly managing the active/inactive state of nodes connected to its own host and other hosts. A node activation status management method. 2. Each of the hosts further includes a mechanism for mutually exchanging management information indicating the active/inactive state of the nodes before communication between the hosts becomes possible. 2. The node activation state management method described in item 1. 3. After each host becomes able to communicate with each other, when there is a change in the active/inactive state of a node connected to each host, the host shall notify other hosts of this status. 3. The node active state management system according to claim 1, further comprising a mechanism for reporting changes in state. 4. Each host further includes a mechanism that, when receiving a report of a change in the state of the node, reflects the state change information of this node in management information indicating the active/inactive state of the node managed by the host itself. 4. The node activation state management method according to claim 3, wherein the node activation state management method is characterized in that: