JP4198217B2 - Operation system and congestion control method in the operation system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a congestion control technology for an operation system including a general-purpose computer that manages a plurality of communication devices via a communication medium, and in particular, dynamically by a plurality of autonomous messages issued from managed communication devices. The present invention relates to an operation system that performs congestion regulation and a congestion regulation method thereof.
[0002]
[Prior art]
In recent years, with the diversification of networks, there has been a demand for an operation system capable of centrally managing a large and complex network. In such an operation system that requires large-scale aggregation, it is necessary to increase the number of management targets that can be managed by a single operation system. In that case, in general, a large number of managed network components (for example, Many messages (static messages such as equipment configuration, mounting information, and dynamic messages such as status change reports and alarm reports) sent from the exchange must be processed in real time, increasing the load on the operation system. However, there arises a problem that the processing capacity of the general-purpose computer that is a component of the operation system is exceeded. How to reduce such an increase in operation system load is an important research subject, and several research presentations have been made.
[0003]
For example, according to the one described in IEICE Technical Report SSE93-51 IN93-58 CS93-74 (1993-09) PP.77-82 (referred to as Conventional Technology 1), the network configuration is reduced to reduce the load on the operation system. Focusing on the urgency of an event that has occurred in an exchange, which is an element, that is, the importance of a message, the collection of monitoring information by real-time notification and the collection of monitoring information by non-real-time notification are separated according to the importance. Specifically, during normal times, only high-importance messages are reported to the operation system, and low-importance messages are regularly sent during times when the operation system is lightly loaded (for example, at night). The log collection process is started.
[0004]
Also, according to the one described in IEICE Technical Report SSE95-129 (1995-12) PP.25-30 (referred to as Prior Art 2), it is managed in the system management of OSI (Open System Interconnection). A plurality of management programs in the operation system serving as a manager for an agent having a plurality of communication devices targeted for communication individually establish a logical line called an association via a communication medium, and perform communication. In order to monitor the status of the management target in the management agent, the operation system sets the conditions for classifying autonomous messages for each management program in the classifier in the management agent so that only necessary autonomous messages can be received. ing.
[0005]
[Problems to be solved by the invention]
As described above, in an operation system that requires large-scale aggregation, it is necessary to increase the number of management targets that can be managed by one operation system. However, in the conventional technique 1, monitoring is performed by real-time notification according to importance. Separates the collection of monitoring information from information collection and non-real-time notification. In normal times, only high-importance messages are reported to the operation system, and low-priority messages are regularly sent during times when the operation system is lightly loaded. However, the borderline for determining whether the message is a high importance message or a low importance message is statically determined in advance and does not change during driving. It is not possible to dynamically perform congestion regulation control according to.
[0006]
Further, in the above-described conventional technique 2, when the number of management targets is increased, the operation system needs to establish associations with a number of agents to be managed and monitor the state of the communication device. Here, the managed program in the operation system sets the autonomous message type that is issued from one agent to the operation system in order to set the classification condition for receiving only the autonomous message that it desires for each association. Limited. However, the classification condition that can be set for the management agent by the management program cannot limit the number of messages issued per unit time even though the type of autonomous message can be limited. Accordingly, since the number of communication devices to be managed increases as the operation system aggregation scale increases, the operation system is more likely to receive autonomous messages from the management agent that exceed the processing performance of the operation system itself. Become.
[0007]
In this way, when the number of autonomous messages exceeding the processing performance of the operation system itself is received from one or more managed communication devices, the operation system is in a congested state due to resource bottlenecks such as the CPU and memory of the general-purpose computer. This causes problems such as delays in displaying autonomous messages and the inability to input control commands. Also, in a large-scale operation system, even if only a few managed agents / devices cause frequent autonomous messages, the operation system itself is congested and does not generate other autonomous messages. Even the management target agent / device is hindered in its monitoring / control. This situation is the same as in the prior art 1.
[0008]
In the operation system that performs large-scale aggregation, the reception of autonomous messages for the self-defense of the operation system itself is dynamically performed in response to the frequent occurrence of autonomous messages by some managed agents / devices. Although it is desirable not to affect the monitoring / control of other managed agents / devices that have not caused frequent occurrence of other autonomous messages by performing congestion control, the configurations disclosed in the above prior arts 1 and 2 Then it is impossible.
[0009]
In view of the above circumstances, an object of the present invention is to provide an operation system for monitoring / controlling communication devices with large-scale aggregation. Even if several managed agents frequently generate autonomous messages, the operation system dynamically In addition, an operation system that automatically detects and regulates autonomous message transmission from the managed agent itself to regulate congestion of the operation system itself due to frequent autonomous messages without consuming operation system resources. Another object of the present invention is to provide a congestion control method in the operation system.
[0010]
[Means for solving problems]
To achieve the above object, the operation system of the present invention, when the autonomous message is issued to the operating system from the agent with the individual communication device, the individual agents (in FIG. 1 16～18,21～23 26-28) means for measuring the number of autonomous messages issued per unit time (same as message counters 4, 6 and 8), and for agents whose number of autonomous messages issued per unit time exceeds a predetermined threshold. On the other hand, a means for automatically sending a signal specifying a restriction condition for restricting the issuance of an autonomous message based on the importance or urgency of the autonomous message and restricting the issuance of the autonomous message from the agent (same condition control as above) and parts 5,7,9), the elapsed time after issuing from the restriction of certain autonomous message, the agent To said means for transmitting a signal for releasing the restriction of the issuance of the autonomous message (same fractional condition control unit 5, 7, 9) and, a congestion flag indicating whether or not the congestion state, the congestion When the flag is OFF, specify the regulation condition to regulate the autonomous message based on the importance or urgency of the autonomous message by performing the step of measuring the number of autonomous messages issued per unit time of the agent and the subsequent processing The congestion flag is set to ON when sending a signal to be sent, and if the congestion flag is ON, the autonomous message received when the number of queues in the system is less than a predetermined threshold is delivered to the management program, and wherein a discarding congestion flag autonomous messages received when the wait queue of the system is equal to or higher than the threshold to a predetermined To have.
In the congestion control method in the operation system of the present invention, when an autonomous message is issued to an operation system from an agent having a plurality of communication devices, a step of measuring the number of autonomous messages issued per unit time of each agent (Steps 103 to 105 in FIGS. 2 and 3), a step of determining whether or not the number of autonomous messages issued per unit time exceeds a predetermined threshold for each individual agent (step 106). A signal that automatically specifies a regulation condition for regulating an autonomous message based on the importance or urgency of the autonomous message is automatically sent to an agent whose number of autonomous messages issued per unit time exceeds the predetermined threshold. step to limit the autonomous message coming from the agent (the same Step 109, the step 203) of FIG. 4, after a predetermined time has elapsed since the limit autonomous messages coming from the agent to the agent, sending a signal to release the restriction of the previous autonomous message (FIG. 5 Step 304) and a step of measuring the number of autonomous messages issued by the agent per unit time when the congestion flag is OFF, and subsequent processing. The congestion flag is set to ON when sending a signal designating a regulation condition for regulating the autonomous message based on the importance or urgency of the autonomous message, and the system waits when the congestion flag is ON. Deliver the received autonomous message to the management program when the number of queues is less than a predetermined threshold, Waiting queue of the stem is characterized by comprising a discarding congestion flag autonomous messages received not less than a predetermined threshold. In addition, the step of measuring the number of autonomous messages issued per unit time of the agent includes a message count value cyclically assigned to the autonomous message and a time when the autonomous message is received every time the autonomous message is received. Is stored in a table, and a sub-step of estimating the number of autonomous messages issued per unit time based on the table.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic configuration diagram in which the present invention is applied to an operation system including a general-purpose computer that manages a plurality of communication devices.
In FIG. 1, 1 is an operation system that manages a plurality of communication devices, 2 is a general purpose computer, 3 is a communication control unit that communicates with agents 15, 20, and 25 to be described later, and 4, 6, and 8 are each agent to be managed. Counters for measuring the number of messages issued per unit time, 5, 7 and 9 receiving a request from the message counter, and a classification condition control unit for setting a classification condition for an agent to be managed, and 10 for communication A management program 11 having a scenario for monitoring / controlling the device, 11 is a timer management unit for starting the programs and transmitting messages to the programs at predetermined times in accordance with timer registration requests of the respective programs, and 12, 13, 14 Association between operation system and agent, 15, 20, 25 are agents, 16-18 21～23,26～28 communication apparatus respectively managed in the agent 15, 20, 25, 19,24,29 are fractionator for fractionating autonomous message in the agent 15, 20, 25, respectively.
[0012]
FIG. 2 is a message reception number measurement table used for measuring the number of messages per unit time.
Next, a method of using the message reception number measurement table will be described with reference to FIG. First, each time a message is received, the message count value i for that message and the time ti when the message is received are sequentially written in the message reception number measurement table. The message count value i starts from 1 and is at most n (see FIG. 2). When the message count value i in the message reception number measurement table reaches the maximum value n and completes a cycle, the next message starts again from the message count value 1 and overwrites the message reception number measurement table. Before overwriting, index the time when the message count value was written one round before, and the new time (current time) about to be overwritten and its index time (time one round before) The difference is taken and it is determined whether the difference is large enough to receive n messages. By this processing, the number of messages received between the time before the first round and the current time can be known. Thereby, conversely, the number of messages received per unit time can be estimated (= n / (current time-1time before the round)).
[0013]
FIG. 3 is a processing flowchart of the message counter.
In FIG. 1, the management program 10 sets classification conditions for the classifiers 19, 24, and 29 of the agents 15, 20, and 25 using the classification condition control units 5, 7, and 9, and monitors autonomous messages for all agents. 3 is described in detail with reference to FIG. 3.
[0014]
First, when each message counter 4, 6, 8 receives an autonomous message from the corresponding agent 15, 20, 25 (step 101), it determines whether the congestion flag is ON or OFF (step 102).
As a result of the determination, if the congestion flag is OFF, after the message count value is increased (step 103; however, if the message count value exceeds n in FIG. The current time is collected (step 104), and the time corresponding to the key is collected from the message reception number measurement table of FIG. 3 using the previously increased message count value as a key (step 105). Then, it is determined whether or not the time difference between the current time and the collected time is less than the threshold (step 106).
[0015]
If the result of determination in step 106 is that the time difference between the current time and the collection time is equal to or greater than the threshold (step 106: N), the message reception number measurement table column corresponding to the message count value is updated. (Step 107), the received autonomous message is distributed to the management program (Step 108).
If the time difference between the current time and the collection time is less than the threshold as a result of the determination in step 106 (step 106: Y), the congestion flag is set to ON (step 106: Y) as if a message exceeding the threshold was received within the specified time (step 109), an over-threshold notification is issued to the classification condition control unit (step 110).
[0016]
If the result of determination in step 102 is that the congestion flag is ON, the number of waiting queues in the system is collected (step 111), and it is determined whether or not the number of waiting queues is less than a threshold value (step 112). If the number of waiting queues is less than the threshold value (step 112: Y), an autonomous message is distributed to the management program assuming that there is still room (step 113). Conversely, when the number of waiting queues is equal to or exceeds the threshold (step 112: N), the autonomous message is discarded as having no room (step 114).
[0017]
FIG. 4 is a process flowchart of the classification condition control unit when a threshold-over notification is received from the message counter.
When the classification condition control units 5, 7, and 9 receive the threshold value over notification from the message counters 4, 6, and 8, respectively (step 201), the corresponding classifiers 19, The current classification conditions set in 24 and 29 are stored (step 202). After that, the fact that the threshold value is exceeded is set, and a classification condition that strengthens the restriction condition is set for the corresponding agent classifier (step 203). Here, the classification condition in which the restriction condition is strengthened refers to a condition for strengthening restriction on the received autonomous message, for example, by limiting the received autonomous message to one having higher importance or urgency. As described above, the classification condition control unit in FIG. 5 is activated after a predetermined time has elapsed since the classification condition with the stricter regulation condition set in the classifiers 19, 24, and 29 corresponding to the over-threshold. Then, timer registration is performed with respect to the timer management unit 11 (step 204).
[0018]
FIG. 5 is a processing flowchart of the classification condition control unit when the activation request from the timer management unit is received.
When the timer registration to the timer management unit 11 in step 204 of FIG. 4 is completed and a start request is received from the timer management unit 11 after a predetermined time has elapsed (step 301), the number of waiting queues in the system is collected. (Step 302), it is determined whether or not the number of waiting queues is less than a threshold (Step 303).
[0019]
If the result of determination in step 303 is that the number of waiting queues is less than the threshold value (step 303: Y), it is considered that the congestion state has been resolved, and the classification condition stored in step 202 in FIG. The state is returned to the state before setting the classification conditions set in the agent classifier and the regulations are strengthened (step 304), and the congestion flag is turned off (step 305).
If the result of determination in step 303 is that the number of waiting queues is equal to or exceeds the threshold value (step 303: N), it is considered that congestion has not yet been resolved, and the timer management unit again receives a timer for a predetermined time. Registration is performed (step 306).
[0020]
In the above embodiment, the “predetermined time” registered in the timer management unit 11 in step 204 in FIG. 4 and step 306 in FIG. 5 is fixed in advance to an appropriate time that the congestion state fluctuates during normal operation. However, it may be changed according to the situation, for example, by setting a predetermined time longer when the congestion state is not easily resolved.
In the above embodiment, the number of issued autonomous messages per unit time of the communication device is measured based on the message count measurement table storing the message count value and the time when the autonomous message was received. The number of autonomous messages received within the unit time measured by the unit time measurement timer may be directly detected.
[0021]
【The invention's effect】
As described above, according to the operation system and the congestion control method of the present invention, even when several managed agents frequently generate autonomous messages, the operation system dynamically and automatically detects it, By restricting autonomous message transmission from the managed agent itself, it becomes possible to restrict congestion of the operation system itself due to frequent occurrence of autonomous messages without consuming resources of the operation system.
[Brief description of the drawings]
FIG. 1 is a diagram for explaining an embodiment of the present invention.
FIG. 2 is a diagram for explaining a message reception count measurement table per unit time in an embodiment of the present invention.
FIG. 3 is a process flowchart of a message counter in an embodiment of the present invention.
FIG. 4 is a processing flowchart of a classification condition control unit when a threshold over notification is received from a message counter according to an embodiment of the present invention.
FIG. 5 is a processing flowchart of a classification condition control unit when an activation request is received from a timer management unit in one embodiment of the present invention.
[Explanation of symbols]
1: Operation system 2: General-purpose computer 3: Communication control units 4, 6, 8: Message counters 5, 7, 9: Classification condition control unit 10: Management program 11: Timer management units 12, 13, 14: Associations 15, 20 , 25: Agents 16-18, 21-23, 26-28: Communication devices 19, 24, 29: Separator

Claims (3)

In an operation system comprising a general-purpose computer that performs communication via a communication medium with a plurality of agents each having a plurality of communication devices, and manages the plurality of communication devices ,
When the autonomous message to the operating system from the individual agents is issued, and means for measuring the issuance number per unit time of autonomous messages of the individual agents,
For agents whose number of autonomous messages issued per unit time exceeds a predetermined threshold, a signal is automatically sent to specify a regulation condition that restricts the issuing of autonomous messages based on the importance or urgency of autonomous messages. Means for restricting the issuing of autonomous messages from the agent;
Means for transmitting a signal for releasing the restriction on the issuance of the autonomous message to the agent after a predetermined time has passed since the issuance of the autonomous message was restricted;
A congestion flag indicating whether or not a congestion state,
A regulation condition for regulating the autonomous message based on the importance or urgency of the autonomous message by performing the step of measuring the number of autonomous messages issued per unit time of the agent and the subsequent processing when the congestion flag is OFF The congestion flag is turned on when sending a signal designating
When the congestion flag is ON, the received autonomous message is delivered to the management program when the number of waiting queues of the system is less than a predetermined threshold value, and the number of waiting queues of the system is equal to or larger than the predetermined threshold value. An operation system comprising a congestion flag for discarding a received autonomous message. In the congestion control method in the operation system that performs communication via a communication medium with a plurality of agents each having a plurality of communication devices, and that includes a general-purpose computer that manages the plurality of communication devices ,
When an autonomous message is issued from the individual agent to the operation system, measuring the number of issued autonomous messages per unit time of the individual agent;
Determining whether the number of autonomous messages issued per unit time exceeds a predetermined threshold for each individual agent; and
A signal that automatically specifies a regulation condition for regulating an autonomous message based on the importance or urgency of the autonomous message is automatically sent to an agent whose number of autonomous messages issued per unit time exceeds the predetermined threshold. Limiting autonomous messages coming from agents;
Transmitting a signal for canceling the restriction of the previous autonomous message to the agent after a lapse of a predetermined time after restricting the autonomous message coming from the agent;
A congestion flag indicating whether or not a congestion state,
A regulation condition for regulating the autonomous message based on the importance or urgency of the autonomous message by performing the step of measuring the number of autonomous messages issued per unit time of the agent and the subsequent processing when the congestion flag is OFF The congestion flag is turned on when sending a signal designating
When the congestion flag is ON, the received autonomous message is delivered to the management program when the number of waiting queues of the system is less than a predetermined threshold value, and the number of waiting queues of the system is equal to or larger than the predetermined threshold value. A congestion control method in an operation system, comprising: a congestion flag for discarding a received autonomous message. In the congestion control method in the operation system according to claim 2,
The step of measuring the number of autonomous messages issued per unit time of the agent is a table of a message count value cyclically assigned to the autonomous message and a time when the autonomous message is received for each reception of the autonomous message. And a sub-step for estimating the number of autonomous messages issued per unit time based on the table, and a congestion control method in an operation system.

Images