WO2001082535A1

WO2001082535A1 - Device for circular polling/response between snmp manager and agent and its method

Info

Publication number: WO2001082535A1
Application number: PCT/JP2000/002593
Authority: WO
Inventors: Masahiro Koyano; Kenji Inoue
Original assignee: Fujitsu Limited
Priority date: 2000-04-20
Filing date: 2000-04-20
Publication date: 2001-11-01

Abstract

A network management device and its method, especially a device for polling/response between an SNMP manager and an agent and its method by which the quality of network operation and management is maintained without degrading the processing capability of the network devices. The method comprises the steps, executed by an SNMP manager, determining an IP address route of the SNMP agents existing on the route from the SNMP manager to the lowermost management objective SNMP agent, creating a route diagram showing organized and sorted routes for each agent, downloading the route diagram to the objective agent, and polling the SNMP agents lower in level than the SNMP manager, and the steps, executed by the SNMP agent, polling the lower-level agents according to the route diagram when polled by the upper-level agent, and sending the response of the agent itself and the responses from the lower-level SNMP agents.

Description

Specification

TECHNICAL FIELD The present invention relates to a system for circulating polls between NMP manager agents and a Z-response apparatus.

The present invention relates to a network management device and a method for managing network devices (routers, switching hubs, ATM exchanges, etc.), and more particularly to a manager Z agent using SNMP (Simple Network Management Protocol). Polling between networks In the Z response, an SNMP manager / agent that maintains the quality of network operation management without reducing the processing capacity of network devices even during traffic congestion The present invention relates to a polling / response device and a method thereof. Background art

SNMP is a network management protocol proposed by a plurality of network vendors (such as Cisco), and is defined by RFC (Request For Comments), such as 1067. The SNMP manager, which manages the entire network, periodically specifies the IP address of the SNMP agent that manages each registered network device, and performs polling periodically. The specified agent returns a response with the requested content.

The management bucket (polling / response) between the manager and the agent reaches the target agent by passing through a plurality of network devices (routers, etc.) forming a network route. As a result, the number of agents on the network route increases, If the logging cycle becomes shorter, the number of management packets passing through it will increase. As a result, the bandwidth of the line was squeezed and the response of the original business data was adversely affected.

As a conventional technique for solving this problem, there is an improved version of a polling Z response algorithm as described below.

1) SNPMV2's Manager-to-Manager Management Information Base, RFC1451, provides two or more polling-oriented managers and divides the polling range (for example, Tokyo and Osaka). However, this is a technology for cooperatively managing networks (Autonomous Systems) with different management entities from the beginning, and since there are multiple SNMP managers, management is especially large for small and medium-sized networks. There is a problem that becomes too much.

(2) “Real-world evaluation of a dynamic polling cycle control method based on real-time changes in network management information” (Information Processing Society of Japan, 57th Annual Meeting, 1998) While capturing the polling interval value and the line load in real time, if the load is high, the polling is temporarily postponed, and if the load is low, the polling is accelerated, so that the overall load on the network bandwidth is reduced. We propose a method that does not give. This method takes advantage of the fact that network management does not require much real-time performance, but has the problem that the quality of network operation management deteriorates due to the temporary postponement of polling.

(3) The “network management method” (Japanese Patent Application Laid-Open No. Heisei 9-228252) is a modification of the above (2), which also suppresses polling. It has a similar problem.

(4) The “network management information collection method” (Japanese Patent Application Laid-Open No. 11-4222) is a method of changing periodic polling triggered by a trap (Trap) from an agent. However, the network equipment is normal There is no trap when operating, and there is a problem that the effect of reducing traffic is low.

(5) Another new attempt is to apply active networks to network management. This is to incorporate a mechanism for dynamically processing the SNM P bucket as needed in the network equipment. This is an attempt to substitute “special processing” for a specific agent, thereby reducing the number of management packets passing through the entire network. For example, “A Study on the Effect of Applying Active Network Technology to Network Management” (Transactions of the IPSJ 58th (Early 2001) National Convention) is as follows. Kitekutiya has been proposed.

i) Instead of polling all network devices in the same way, execute a network management program on each corresponding node according to the management purpose of each device, such as routers and ATM switches, and change there. A method of notifying the upper SNMP manager only when there is an error. This is a variant of the trap method.

ii) A method of notifying a response to polling but notifying only when there is a change. This is a combination of polling and trapping.

However, the above-mentioned active network includes a network device because a network device brings a packet that should be routed in the network layer to the application layer and processes it. Problems have been pointed out, such as consuming more equipment resources, which in turn causes a reduction in processing capacity. Disclosure of the invention

In view of the above problems, an object of the present invention is to reduce the network traffic. Even if the traffic state changes from a traffic state to a high traffic state, the SNMP manager maintains the quality of network operation management without suppressing polling. An object of the present invention is to provide a polling Z response device and a method thereof.

Further, an object of the present invention is to provide a processing mechanism that, when the network changes from a low-traffic state to a high-traffic state or vice versa, appropriately processes the processing mechanism according to the state after the change. The P manager downloads to the network agent side, and the agent executes the processing based on the processing mechanism. The polling between the SNMP manager / agent and the Z response device and It is to provide the method. Furthermore, an object of the present invention is to relay one or more network devices through which a sent or returned management bucket passes before reaching the target agent or manager. To provide a polling / response device between the SNMP manager and the Z agent configured so as not to impair the performance of the switching / switching function and a method thereof.

According to the present invention, there is provided a polling / response method between an SNMP manager and an agent, wherein one of the SNMP managers goes from an SNMP manager to a managed SNMP agent at the lowest layer. Calculating an IP address route of an agent that passes through the above, generating a route table in which the IP address route is sorted and classified according to the route for each of the agents: Downloading to the target agent, polling the SNMP agent below the SNMP manager, and downloading the other SNMP agent to the target agent. Based on the routing table, when polling from the higher level, it also polls the lower-level agent and returns its own response and the response received from the lower level. A method for polling / responding between SNMP managers / agents is provided.

Further, according to the present invention, there is provided a polling response method between the SNMP manager / agent, wherein one of the SNMP managers executes a processing program registered in advance in the manager. Transferring to the agent the extended MIB of the target agent, transmitting the initial data of the processing program to the agent, polling the agent, and The SNP agent generates a task process according to the transferred processing program, enters a process waiting state, starts the task process generated by receiving the initial data, and executes the port process. Performing a task process started when a ring is received, the method comprising: a polling Z response method between the SNMP manager Z agents. You. Further, according to the present invention, there are provided an SNMP manager device and an SNMP agent device corresponding to the above method.

Based on the above configuration, the number of network devices managed by polling between the SNMP manager / agent can be reduced without reducing the number of network devices and without increasing the polling time for management / monitoring. The number of management buckets can be reduced. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a diagram showing a basic configuration example of an SNMP manager according to the present invention.

FIG. 2 is a diagram showing a basic configuration example of the SNMP agent according to the present invention.

FIG. 3 is a diagram illustrating an example of a network configuration.

KI 4 is a diagram showing an example of a route to a target agent. FIG. 5 is a diagram showing an example of the routing table.

FIG. 6 is a diagram showing an example of generation of a processing task in the SNMP agent.

FIG. 7 is a diagram showing an example of normal polling.

FIG. 8 is a diagram showing an example of circulation polling according to the present invention. Figure 9 shows an example of a single response and a concatenated response.

O o

Figure 10 shows an example of the format of a consolidated response frame.

FIG. 11 is a diagram illustrating an example of the abnormal trap process.

FIG. 12 is a diagram showing the relationship between the abnormal trap and the polling target again.

FIG. 13 is a diagram showing an example of a routing information collection sequence by normal polling.

FIG. 14 is a diagram showing an example of an agent processing task activation sequence.

FIG. 15 is a diagram showing an example of a sequence of the cyclic polling and the response.

FIG. 16 is a diagram showing an example of the abnormal trap sequence. FIG. 17 shows an example of the end sequence of the cyclic polling.

FIG. 18 is a diagram showing an example of the state transition of the SNMP manager. ■

FIG. 19 is a diagram showing an example of the state transition of the SNMP agent. BEST MODE FOR CARRYING OUT THE INVENTION FIG. 1 shows an example of a basic configuration of an SNMP manager according to the present invention.

The white block on the upper side of the SNMP Manager 1 indicates a conventional processing block, and the polling issue processing unit 13 uses the polling table 14 to lower the lower agent. The response is processed by the response reception processing unit 11. On the other hand, the shaded block under SNMP Manager 1 shows a block for processing the present invention. The start processing unit 15 performs start processing of the SNMP management network, and at that time, the routing table creation processing unit 17 creates a routing table used for cyclic polling according to the present invention, and Stored in group 21. The routing table / processing program transfer processing unit 20 transfers the corresponding routing table and processing program 22 to the agent. Further, the abnormal trap reception processing unit 18 receives the Trap command from the agent, and accordingly, the polling 0 n Z 0 ff issuance processing unit 19 responds to the failure trap. Issuance of polling to Z is controlled, and if necessary, the termination processing unit 16 declares the end of cyclic polling. The detailed operations will be described later.

FIG. 2 shows a basic configuration example of an SNMP agent according to the present invention.

Again, the white block above the SNMP agent 2 indicates a conventional processing block, and the response creation processing unit 32 receives a poll from the SNMP manager 1 and responds to the response. Create a connection. On the other hand, the shaded block under SNMP Agent 2 shows a block for processing the present invention. The start processing unit 33 starts the corresponding processing task by the processing program downloaded from the management information to the extended MIB (Management Information Base) and circulates by receiving the routing table. Start polling You. The polling processing unit 35 performs cyclic polling using the polling table 38 created based on the received routing table 39. The response link processing unit 36 links its own response when polled from the upper level with the response received first from the lower level and returns it to the upper level. When detecting the abnormality of the lower agent, the abnormality trap processor 37 raises the Trap command to the higher level, and further excludes the agent from the target of the cyclic polling. The termination processing unit 34 terminates and stops the polling in response to a circulation polling termination declaration from the SNMP manager.

FIG. 3 shows an example of a network in which the present invention is executed. Here, first, the following processing is performed in the start processing unit 15 and the routing table creation processing unit 17 of the SNMP manager 1 shown in FIG.

When SNMP is started in the network shown in FIG. 3, the start processing unit 15 of the SNMP manager of the IP (Internet Protocol) address M normally starts network management by polling, and the polling is started. It measures the return time of the response to the response, the number of agents in the response after the time-out, and the like. As a result, when it is determined that the lines have reached the specified threshold value and the line has become congested, the routing table creation processing unit 17 of the SNMP manager enters a process of creating a “routing table”. Here, the route indicates how the SNMP management packet (polling or response) flows through the network device by the IP address A, An, B, or Bn of the agent. It is a thing. For example, in the example of FIG. 3, as shown by the dotted line, when polling the agent of the SNMP manager IP address A3 by normal polling, the network device A—A1 is used. . This route is determined by the route determining mechanism of the routers A and A1, which are the network devices, and the route determining mechanism of the switch. The response is reversed. Note that this TCP / IP commands such as ping, traceroute, and ns1ookup are used as IP route search tools to find such routes.

FIG. 4 shows an example of a route to the target agent shown in FIG.

The left column of FIG. 4 shows the IP addresses A, An, B, and Bn of the target agent, and the right column shows the IP addresses of the passing agents. The above example is indicated by a dotted frame. This route information is retained as original registration information by the SNMPP manager ο

FIG. 5 shows an example of a routing table created by the routing table creation processing unit 17 of the SNMP manager according to the present invention.

In the routing table of FIG. 5, the IP address M of the SNMP manager and the IP addresses A, Α1,... Of the adjacent agents are sequentially cut out from the route to the target agent shown in FIG. Things. In the left column, the IP address of the own device is described, and in the center column, the IP addresses from the S NMP manager's IP address M to the own device and the lower device directly under the IP address of the own device. The IP address is described. In the right column, the IP address of the lower-level device to be subjected to cyclic polling (to be described later) by the own device is described along with polling target parameter information. In this way, the above routing table is copied from the SNMP manager's original registration information together with the information on what kind of polling parameters are to be polled for an agent. . The route table data created for each agent is stored in the route table group 21.

As will be described later, by holding down the above route, the superimposed multiple management buckets are reduced to the minimum required number of management buckets, and furthermore, Since the management packet itself passing through the agent of the network device passing through has a meaning, processing at the network layer becomes possible, and processing delay at the application layer in the active network architecture described above The problem is solved. FIG. 6 shows an example of processing task generation in the SNMP agent according to the present invention.

For this, the following processing is executed between the route table / processing program transfer processing unit 20 of the SNMP manager and the start processing unit 33 of the SNMP agent. In the present invention, an SNMP extension MIB 40 (in this example, a boot), which is an extension of the MIB (Management Information Base), which is a standard database of the agent, is provided on the side of the network device to be the agent. The processing program downloaded from the SNMP manager is registered in it.

The routing table of the SNMP manager and the processing program transfer processing unit 20 correspond to the processing program (a cyclic polling program to be described later) from the database 22 in which the processing program to be executed on the agent side is registered in advance. And / or reads the corresponding agent routing table from the routing table group part 21 described above, and uses the SMP basic command Set-Request to read the agent's SNMP extension MIB boot. Set it to On the agent side, when all the contents of the processing program are set in the boot, the start processing unit 33 generates a corresponding processing task from the set processing program, and then waits for processing. Becomes With this mechanism, the agent 3 can take in the processing corresponding to the various processing programs to be set even during the operation management operation.

Next, circulation polling according to the present invention will be described.

Fig. 7 shows an example of conventional normal polling. FIG. 8 shows an example of the cyclic polling according to the present invention. First, the conventional normal polling shown in FIG. 7 will be briefly described. In the normal polling, in the SNMP manager of the IP address M, Polling issuance processing unit 13 The IP address A, An, B, and Bn are sequentially polled for each agent. Then, the polled agent returns the response created by the response creation processing unit 32 to the manager each time. The manager performs the processing in the response reception processing unit 11.

On the other hand, in the cyclic polling according to the present invention shown in FIG. 8, the SNMP manager side route table / processing program transfer processing unit 20 and polling on / off issue processing unit 19, and the SNMP agent 3 The following processing is performed by the polling processing unit 35 and the response connection processing unit 36 on the side.

As described above, after the creation of the routing table, the SNMP manager uses the routing table and the processing program transfer processing unit 20 to execute all the management agents X at a certain timing, for example, when the time is specified or the threshold is exceeded. The routing table corresponding to each is transmitted. For example, to the agent at the IP address A1, only the data of the item A1 in the routing table of FIG. 5 is transmitted using the SNMP basic command Set-Request. This causes the SNMP manager to declare the “start of cyclic polling” according to the present invention. In the following cyclic polling, the SNMP manager polls only the agent of the IP address A and B immediately below it. Accordingly, the polling 0 n Z 0 ff issuance processing unit 19 turns off the polling request for the agents of the other IP addresses A 1 to A 7 and B 1 to B 5 in the polling table. . 'On the other hand, each agent receiving the routing table from the SNMP manager The following processing is performed by the polling processing unit 35 and the response connection processing unit 36. The polling processing unit 35 knows the IP address of the next agent to be polled and the information of the parameter to be polled from the routing table unit 39 that stores the received routing table. Then, a polling table 38 to be executed by itself is created. For example, in the case of the agent of the IP address A1, the polling targets are the agents of the IP addresses A3, A4, and A5.

When each agent receives a poll from above, it creates its own response as usual. However, according to the cyclic polling of the present invention, the response is not immediately returned but is temporarily stored. At the same time as receiving the polling, it polls the lower-level agent based on the contents of the polling table 38 (circular polling). For example, as shown in FIG. 8, the agent of the IP address A issues two polls A → A1 and A → A2 to each agent branching under the IP address A. Similarly, the agent at IP address A1 issues three polls A1 → A3, A1 → A4 and A1 → A5 to each agent under the IP address A1. .

FIG. 9 shows an example of a single response and a linked response in the circulating polling of the present invention. Here, the response connection processing unit 36 performs the following processing.

In the example of FIG. 9, each of the agents at IP addresses A 3, A 4, and A 5 is transmitted from the agent at IP address A 1 because no agent exists below itself. Upon receiving the circular polling, it returns its response immediately as usual (single response). The agent of IP address A1 receiving the response first sends the IP address first. A's own response that was temporarily held when cyclic polling was received from the agent of A, and the lower-level IP address A response (consolidated response) obtained by combining the response received first from the response A3, A4, and A5 agent is returned to the agent of the upper IP address A.

In the example of FIG. 9, the agent of IP address A1 receives the first response from the lower-level agent (the response from the agent of IP address A3) and its own response (A1). ) Is returned with a concatenated response (A 3 + A 1). Similarly, the agent of IP address A, which receives normal polling from the SNMP manager and cyclically polls each of the lower-level IP addresses A1 and A2, sends the first received IP address. The response (A 3 + A 1) from the agent at address A 1 is connected to the own response (A), and the connected response (A 3 + A 1 + A) is sent to the SNMP manager. I will send it back.

Figure 10 shows an example of the frame format of the consolidated response.

(A) of FIG. 10 shows a normal SNMP frame format, and (b) shows a configuration of the SNMP message field. These are general and will not be described here. FIG. 10 (c) shows an example of the connection response of the present invention arranged in the bucket data unit (PDU). The PDU section is divided and concatenated into a plurality of Get-Response messages, and each Get-Responsel, Get-Response 2,... A) and other responses (A3, A1, A).

As described above, in the concatenated response of the present invention, in order to minimize the response delay for polling from the upper level, only the response received first from the lower level is concatenated with its own response and returned. Send. An agent that has already returned its own response immediately returns the destination address of the response that has risen from the lower level to the IP address M of the SNMP manager, and immediately returns the response. For example, if the agent at IP address A1 returns a concatenated response (A3 + A1) including its own response, and then receives a response (A4) from the agent at the lower IP address A4. The response is returned to the SNMP manager as a response (A4) with the destination IP address set to M.

The SNMP manager receives the concatenated response and the single response of all the agents (A, A1, A2, A3) on a certain route and sends them to the SNMP application as usual. And reflect it in network management.

As described above, according to the cyclic polling of the present invention, the polling function, which has been conventionally managed by the SNMP manager, is distributed to the network devices on the branch route, so that each route can be assigned to each route. The problem that the management bucket is passed in duplicate is eliminated. Furthermore, the response with the fast response is bundled by using the concatenated response, while the response with the slow response is returned to the SNMP manager as it is. As a result, the number of responses can be reduced and the response delay can be prevented at the same time.

Next, a trap (Trap) process performed when an abnormality occurs in the network device during the above-described circulation polling will be described.

This trap processing is performed by the SNMP trap manager 18 and the polling 0 n / off issue processor 19 of the SNMP manager, the polling processor 35 of the SNMP agent, and the response. This is performed by the connection processing unit 36 and the abnormal trap issue processing unit 37.

FIG. 11 shows a network device failure at IP address A3, a path failure between the agents at IP addresses A1 and A3 (circuit failure), and Indicates that a route change or the like has occurred during that time.

First, the operation on the agent side will be described. The agent at the IP address A1 returns the response from the agents at the IP addresses A4 and A5 received by the cyclic polling to a higher order as a concatenated response or the like as described above. On the other hand, for the agent at IP address A3 that has no response even after waiting for a certain period of time, in order to avoid an erroneous failure determination, a predetermined number of re-pollings using the polling response wait time Do. As a result, if the response is normally received, the own response has already been returned. Therefore, the IP address of the received response is replaced with the IP address M of the SNMP manager, and the response is directly returned to the host.

If there is no response even after the predetermined number of re-pollings, only the agent at IP address A1 can know the agent failure of IP address A3 by cyclic polling. Then, the abnormal trap issue processing unit 37 informs the SNMP manager of the fact using the trap of the SNMP basic command, and disconnects the agent of the IP address A3 from the routing table 39. This stops the polling of the IP address A3 agent.

If the response does not increase from all the lower-level agents for a predetermined time after the execution of cyclic polling by the agent at IP address A1, a failure or polling / response error will occur. Judge that a response has occurred, give up the connection of the response, and immediately transfer only the own response that has been temporarily suspended to the IP address M of the SNMP manager and return it. After that, re-polling is performed a predetermined number of times using the polling / response waiting timer. Subsequent operations are the same as above.

Next, the operation of the SNMP manager will be described. When the SNMP trap manager's error trap reception processing unit 18 receives the error trap, the polling on / off issue processing unit 19 sends the corresponding polling request in the polling table 14. turn on. As a result, in the example of Fig. 12, the agent of IP address A3, which has been separated from the target of cyclic polling by the agent of IP address A1 once, becomes the target of polling of the manager itself this time. It is returned and its recovery is continuously monitored (see dash-dot line in Figure 11).

Note that the SNMP manager determines that it is better to recreate a new polling route if the number of such traps increases and the polling target reaches a certain ratio or more. As a result, the termination processing unit 16 once declares the “end of cyclic polling”. Then, the management network is rebuilt and re-executed by the creation of the routing table mentioned earlier.

Fig. 12 shows the agent from which an abnormal trap is detected, the abnormal trap parameter (failure location) included in the agent, and the target for re-incorporating the polling of the SNMP manager that received it. It shows the relationship with the agent.

In the example described above, the parameters of the Trap command received from the agent of the I address A1 include the IP address A3 of the agent in which the abnormality is detected. Therefore, the SNMP manager sets the agent of the IP address A3 as a target to be included in the polling again. Trap commands from other agents (Α2, Β2, · ■ ·) are processed in the same way.

FIGS. 13 to 17 show the individual processes described above in a sequence chart. FIG. 18 shows them in the form of state transition of the SNMP manager, and FIG. 19 shows them in the form of state transition of the SNMP agent. Figures 13 to 1 are shown here for easy understanding. The sequence chart in Fig. 7 will be mainly described, and the state transitions in Figs. 18 and 19 will be described with the step numbers in Figs. 13 to 17 in the corresponding places in order to avoid redundant description.

Figure 13 shows an example of a collection sequence of route information by normal polling. When the SNMP of the network is started, the SNMP manager usually performs network management by polling (S11 and S13) for a while, and then responds to the responses (S12 and S14). Create routing table 21 from

FIG. 14 shows an example of an activation sequence of an agent processing task. The SNMP manager sets the processing program 22 to the extension MIB (boot) 40 of the agent using the SNMP basic command Set-Request (S21 and 23). As a result, the corresponding processing task is started on the agent side whose contents have been updated, and the agent enters the processing waiting state.

Figure 15 shows an example of cyclic polling and its response sequence. At a certain moment, the manager sets the routing table corresponding to all the management agents by using the SNMP basic command Set-Request (S31 and S33). This initiates circular polling. Upon receipt of the routing table from the SNMP manager, the agent knows the IP address of the agent to be polled next and information on the parameters to be polled. Upon receiving a poll from the host (S35), create your own response as usual and temporarily hold it. At the same time, cyclic polling is performed at the lower level (S36), and the response received first (S37) and the own response are connected and returned to the higher level (S38). .

Figure 16 shows an example of an abnormal trap sequence. If a line failure or the like occurs on the lower agent (X + 1) side, If the response does not increase from the lower agent (X + 1) even after the agent (X) performs cyclic polling and waits for a certain time (S42 and T. 0.), the self The agent holding the response recognizes that a failure or a polling Z response error has occurred, immediately gives up the connection, and returns the held response by replacing the IP (S43). However, it is possible that the polling Z response has simply been lost, so the same agent is polled again (S44). If you receive the response normally, your response has already been returned. Polling-If the response wait timeout (T.0.) Is repeated the specified number of times and there is no response and there are no unprocessed agents, it is assumed that an error has occurred in the lower-level agent. It is given to the SNMP manager by using the trap (S45). Then, the route table in that direction is separated (S46). After that, the SNMP manager normally performs polling for the failure agent (X + 1) (S47).

Figure 17 shows an example of the end sequence of cyclic polling. The SNMP manager determines that it is better to re-create a new polling route when the polling target of the SNMP manager reaches a certain ratio or more by receiving the Trap command, Declare “end of cyclic polling” by using the Set-Request command with the parameter set to “end” (S51 and 53). As described above, the present invention employs the following characteristic mechanisms.

(1) A mechanism in which each agent polls a lower-level agent and returns a response to a higher-level route in the opposite route.

(2) When receiving SNMP bucket addressed to itself, only own response In the SNMP application layer.

With the above two mechanisms, the SNMP manager can maintain the number of SNMP agents (network devices) managed by cyclic polling without reducing the number of SMP agents (network devices) (there is no need for division management by multiple managers). The adoption of a concatenated response can reduce the number of management packets on the route without increasing the management / monitoring polling time.

As a result, the line bandwidth at the center where the SNMP manager is installed is significantly reduced, and sufficient congestion tolerance is compensated. In addition, as a result of the distributed network management, the use of public WAN (Wide Area Network) lines will also reduce WAN line charges. Furthermore, since the processing load of the SNMP manager is reduced, the number of agents to be managed can be increased more than before.

Claims

The scope of the claims

1. A polling response method between SNMP managers / agents,

S NMP Manager

Obtaining the IP address path of the agent from the SNMP manager to the lowest managed SNMP agent,

Generating a routing table in which the IP address routes are arranged and classified for each of the agents;

Downloading the routing table to a target agent;

Polling the S NMP agent below the S NMP manager;

The S NMP Agent

Based on the down-listed routing table, when polling from the upper level, poll the lower-level agent, and return its own response and the response received from the lower level. A polling / response method between SNMP managers and Z agents characterized by comprising

2. The SNMPP agent further

Temporarily suspending its own response, linking it with the earliest of the responses received from the lower order, and returning it as one response;

For the response received from the lower level after returning its own response, the destination address is replaced with the IP address of the SNMP manager and returned.

The method of claim 1 comprising:

3. The SNMP agent further

When there is no response from the lower order for a predetermined time, the address of the own response is replaced with the IP address of the SNMP manager and immediately returned.

3. The method of claim 2, comprising:

4. The SNMP Agent further

If there is no response even after repeating the polling a predetermined number of times for the lower order without the response, notify the SNMP manager of the abnormality of the lower-level agent with no response. Excluding lower-level agents without polling,

4. The method according to claim 3, comprising:

5. The SNMP manager further comprises:

By receiving the notification, setting an agent without the response as a target of the polling;

5. The method according to claim 4, comprising:

6. The SNMPP manager further

Notifying the target agent that downloaded the routing table of the end of polling based on the routing table;

The method of claim 1 comprising:

7. A polling / response method between SNMP managers / agents.

S NMP Manager

By transferring a processing program registered in the manager in advance to the extended MIB of the target agent that executes the processing program,

Sending initial data of the processing program to the agent Polling the agent;

The S NMP Agent

Generating a task process by the transferred processing program and entering a process waiting state;

Starting the generated task processing by receiving the initial data;

Performing the started task processing upon receiving the polling. The method of claim 3, further comprising: performing a polling Z response method between the SNMP manager / agent.

8. From the IP address route of the agent passing from the S NMP manager to the managed S NMP agent at the lowest layer, the routing table organized and sorted according to the route for each agent described above is created. A routing table creation processing unit to be created;

An SNMP comprising: a routing table transfer processing unit that transfers the routing table to a target agent; and a polling issue processing unit that performs polling to an SNMP agent directly below the SNMP manager. Manager and

A polling processing unit that, based on the transferred routing table, performs polling for a directly lower-level agent when polling from the directly higher level;

A polling Z response device, comprising: an SNMP agent device having: a response processing unit that returns its own response and the response received from the lower order.

9. The response processing section of the SNMPP agent device further includes Holds its own response temporarily, concatenates it with the earliest of the responses received from the lower order, returns it as one response, and returns it after receiving its own response. 9. The apparatus according to claim 8, wherein, for the response, the destination address is replaced with the IP address of the SNMP manager apparatus, and the response is returned.

10. The response processing unit of the SNMP agent device further includes

10. The method according to claim 9, wherein when there is no response from the lower order for a predetermined time, the destination address of the own response is replaced with the IP address of the SNMP manager device and immediately returned.

1 1. The SNMP agent device further includes

If there is no response even after repeating the polling for the lower order without the response a predetermined number of times, an error in the lower-level agent device with no response is notified to the SNMP manager, and the response of the response is notified. 10. The method according to claim 10, further comprising: an abnormality notification processing unit that excludes a lower-level agent device that is not subject to the polling.

1 2. The SNMP manager device further

21. The method according to claim 11, further comprising: a polling issue processing unit that targets the agent device that does not respond to the polling by receiving the notification.

1 3. The SNMP manager device further

9. The method according to claim 8, further comprising a termination processing unit that notifies a target agent device that has downloaded the routing table of a polling termination based on the routing table.

1 4. A processing program that transfers a processing program registered in advance to the manager to the extended MIB of the target agent that executes the processing program. An initial data transfer processing unit for transmitting initial data of the processing program to the agent;

A polling issue processing unit for polling the agent, an SNMP manager device comprising:

A task generation processing unit that generates task processing according to the transferred processing program, and is in a processing waiting state;

A task start processing unit that starts the generated task processing by receiving the initial data;

A polling processing unit that performs the task processing that has been started when the polling is received, and a SNP agent device having:

A polling / response device characterized by comprising:

Fig.1