CN1901568B - A method for realizing historical performance collection in a network management system - Google Patents

Abstract

The invention discloses a method for realizing historical performance collection in a network management system. Customize the maximum number of performance collection threads that can be created and started by the network management system through configuration files, define different performance collection thread classes for different types of network elements, and customize the maximum number of network elements for each performance collection thread according to different network element types. When a network element is added or deleted, after receiving the corresponding message, the performance acquisition control thread automatically assigns the newly added network element to a certain or creates a new thread to collect historical performance of the network element, or automatically traverses each thread until it is found The thread containing the deleted network element deletes the network element from the collection network element list of the thread. When the network management system is started, the network element status polling thread is started, which is responsible for monitoring the changes of the network element status. The invention provides high-efficiency, high-expandability and high-reliability historical performance collection, and is easy to adapt to the requirements of a large-scale network management system with mixed management of various network elements.

Description

A method for realizing historical performance collection in a network management system

technical field

The present invention relates to a network management system, in particular to a method for realizing historical performance collection in a network management system, and in particular to a method for realizing high-efficiency and large-scale historical performance collection in a network environment with different types of network elements and numerous network element devices method.

Background technique

Performance management is an important function of a telecom network management system, which includes collecting statistics, maintaining and checking history of system status.

In the existing system, the first method for performance management is that the user manually starts a performance collection task, and the user specifies the range of network elements to be collected, the performance amount to be collected, the collection time interval, and the collection time. After the performance collection task is successfully created, the network management system starts a performance collection thread, and periodically collects historical performance data from network elements according to the parameters passed in by the created performance collection thread, and provides historical performance data required for daily operation and maintenance to network management operators.

The disadvantage of this method is that the historical performance data can only be obtained after manually starting the performance collection task. In the actual operation process, if the number of network elements in the network is large, the user will not Both create a performance collection task, so that when a performance failure occurs on a network element, the user needs to analyze the cause of the performance failure based on historical performance data, but finds that the network element has no history because no performance collection task has been created for the network element. Performance data can be used for performance failure analysis.

The second method for performance management is to obtain a list of different types of network elements from the database after the network management server is started, and then create a performance collection thread for each network element, and each performance collection thread periodically goes to this network element for performance collection. For a newly discovered network element, a collection thread is automatically created for the network element. After the network element is successfully deleted, the collection thread corresponding to the network element is automatically stopped and deleted.

The disadvantage of this method is that when the number of network elements is relatively large, creating a thread for each network element will occupy a large amount of system resources, and when there is a problem with the communication link between the network element and the network management system, the thread cannot be stopped in time. The performance collection of that network element will also reduce the efficiency of performance collection.

Contents of the invention

The technical problem to be solved by the present invention is to provide a method for realizing high-efficiency and large-scale historical performance collection in a network management system, which can improve the efficiency and speed of collection through technologies such as multi-threading and thread pool.

The present invention is specifically realized like this:

A method for realizing historical performance collection in a network management system, comprising the following processing:

In the first step, when the network management server is started, the performance acquisition control thread is started;

The second step is to obtain various historical performance collection configuration parameters from the configuration file of the performance collection control thread;

The third step is to implement a dynamically loadable interface for each performance acquisition thread;

In the fourth step, in the performance acquisition control thread, according to the network element types managed by the network management system, a list of each network element is obtained through multiple database queries;

The fifth step is to dynamically create a performance collection thread of this type of network element according to the various historical performance collection configuration parameters obtained from the configuration file of the performance collection control thread for the obtained list of certain network elements;

In the sixth step, in the performance collection thread corresponding to a certain network element, each time the timing period arrives, the performance collection is performed cyclically for each network element in the collection network element list of the performance collection thread.

The acquisition of various historical performance acquisition configuration parameters from the performance acquisition control thread includes performance acquisition threads related to network element types, the maximum number of acquisition threads for each type of network element, and the maximum number of acquisition network elements for each type of acquisition thread.

Before collecting the performance of a certain network element, obtain the state of the network element through the SNMP protocol;

If the acquisition of the network element status fails, delete the network element from the collection network element list in the performance collection thread;

If after deleting the network element, there are no other network elements in the collection network element list of the performance collection thread, stop the collection thread and delete it.

When the network management system finds a network element, it sends a request to the performance collection control thread to start network element performance collection. The performance collection control thread searches for all performance collection threads that have been started for this type of network element, and then judges whether the network element already exists. In the collection network element list of a certain performance collection thread;

Return if it exists, otherwise, add this network element to the collection network element list of a performance collection thread whose number of collection network elements does not reach the configured maximum number of collection network elements per thread;

If the number of collection network elements in all performance collection threads currently created has reached the configured maximum value, but the total number of collection threads has not reached the maximum allowable number of performance collection threads, create a new performance collection thread for this network element .

When the network management system deletes a network element, it sends a request to the performance collection control thread to stop the performance collection of the network element. The performance collection control thread searches all the performance collection threads that have been started for this type of network element, and finds the performance collection thread where the network element is located. thread, delete this network element from the collection network element list, if after deleting this network element, the collection network element list of this performance collection thread no longer contains other network elements, stop this collection thread and delete it.

In the method for realizing historical performance collection in the network management system, when a large number of network elements fail to communicate with the network management system, the first step further includes:

When the network management server is started, all network elements are obtained from the database;

Obtain the maximum number of simultaneous network elements collected by each polling thread from the polling configuration file;

Start a cyclic status polling thread for multiple network elements one by one.

When the status polling thread finds a network element, it sends a request to the performance collection control thread to start the performance collection of the network element. The performance collection control thread searches all the performance collection threads that have been started for this type of network element, and then judges whether the network element is Already exists in the collection network element list of a certain performance collection thread;

Return if it exists, otherwise, add this network element to the collection network element list of a performance collection thread whose number of collection network elements does not reach the configured maximum number of collection network elements per thread;

If the number of collection network elements in all performance collection threads currently created has reached the configured maximum value, but the total number of collection threads has not reached the maximum allowable number of performance collection threads, create a new performance collection thread for this network element .

When polling the network element fails, send a request to stop the performance collection of the network element to the performance collection control thread, and the performance collection control thread searches all the collection threads that have been started for this type of network element, and finds the performance collection thread where the network element is located. Delete this network element from the collection network element list. If after deleting this network element, the collection network element list does not contain other network elements, stop and delete this collection thread.

The network management system discovers or deletes a network element successfully, and sends a message to the status polling thread to start or stop the status polling of the corresponding network element.

The method of the invention can provide a high-efficiency, high-expandability, and high-reliability historical performance collection function, and can easily adapt to the requirements of a large-scale network management system with mixed management of various network elements.

Description of drawings

Figure 1 BWA unified network management system performance collection flow chart;

Fig. 2 is a flow chart of newly discovering a network element;

Fig. 3 is a flow chart of deleting an existing network element;

Fig. 4 is a flow chart of processing when a network element polling thread finds that a certain network element is turned from unavailable to on;

Fig. 5 is a flow chart of processing when a network element polling thread finds that a certain network element is turned from passable to unreachable.

Detailed ways

Taking the network management system of the Broadband Wireless Access (BWA) system as an example, the implementation method of the present invention will be introduced in detail below.

The broadband wireless access system (BWA) provides telecom operators with high-speed, large-capacity, high-reliability, and full-duplex broadband access methods in a point-to-multipoint signal transmission mode, and realizes broadband wireless access from user terminals to the backbone network. access. The BWA network management system is a subsystem of the BWA system. In this embodiment, the BWA network management system is a standard client/server structure, the background database adopts Microsoft SQL Server, and the operation and maintenance personnel use a graphical user interface (GUI) terminal to access the SQL Server database through the BWA network management server. The network management server communicates with the network element system through the SNMP protocol.

As shown in accompanying drawing 1, 1, start a performance acquisition remote service (RMI: Remote Method Invocation) in the network management server start-up process. Create a performance collection control thread in the performance collection remote service;

2. The performance acquisition control thread obtains the performance acquisition parameters from the performance acquisition configuration file, firstly reads the performance acquisition parameters, including whether to perform performance acquisition, the time interval of performance acquisition, the class name of the performance acquisition thread related to the network element type, each The maximum number of collection threads of network elements and the maximum number of collection network elements of each collection thread.

Each performance acquisition thread class implements an interface, which can be dynamically loaded. In this way, if a new type of network element is added, only a new collection thread class for this network element needs to be added, and then the configuration file for this network element Metatype configures the newly added acquisition thread class, and does not need to modify the performance acquisition control thread class.

3. After obtaining the performance acquisition parameters, in the performance acquisition control thread, obtain all network element types of the BWA series from the performance acquisition configuration file in the BWA network management system database, and then obtain each type of network element through database query operations list;

4. The performance collection control thread creates a performance collection thread based on the performance collection parameters and the network element list, and allocates some network elements for each performance collection network element thread. The number of network elements in the collection network element list of the performance collection thread cannot exceed the configuration file The maximum number of acquisition network elements per thread set in ;

5. Each performance collection thread automatically starts a timer during the creation process, and the timing period is the performance collection time interval. Whenever the timing period of the performance collection thread arrives, perform the following operations in a cyclic sequence on each network element in the collection network element list:

Firstly, judge whether the SNMP link between the NE and the network management system is normal. If not, delete the NE from the collection NE list. If the NE is deleted, there is no For other network elements, the collection thread is stopped and deleted, and a request to stop performance collection of the network element is sent to the performance collection control thread. Then perform the performance collection operation on the next network element;

If the SNMP link between the NE and the network management server is normal, use the SNMP protocol to obtain performance data from the NE side and write it into the network management database. Then perform the performance collection operation on the next network element until all the performance collection operations on all the network elements in the collection network element list are completed.

The above is the main process of performance collection. In order to improve the efficiency of performance collection and meet the needs of network changes, the performance collection control thread can receive external messages to start or stop the performance collection function of a certain network element.

The network elements in the network management system can be dynamically added and deleted, and the state of the network elements may also change. The present invention provides the processing for these two situations.

The processing flow of adding a network element is shown in Figure 2. After the network management server finds a network element successfully, it sends a message to the performance collection remote service, requesting to start the performance collection function of the network element. After receiving the request message, the performance collection remote service forwards the message to the performance collection control thread. After the performance acquisition control thread receives the request message, it searches for all created acquisition threads. For each collection thread, it is judged whether the network element requesting to start the performance collection function is included in the collection network element list. If the network element requesting to start the performance collection function has been included in the collection network element list of a certain collection thread, the request response ends. If the network element requesting to start the performance collection function does not exist in the collection network element list of any started collection thread. The performance collection control thread starts to start the performance collection function of the network element. The performance acquisition control thread searches all created acquisition threads, and searches for threads that have not yet reached the maximum number of acquisition network elements. The performance collection control thread adds the network element to the collection thread, and the request ends. If all the collection threads that have been created have reached the maximum number of collection network elements, and the total number of collection threads has not reached the maximum number of collection threads, the performance collection control thread creates a new collection thread and adds the network element to the collection thread middle. End of request.

The processing flow of deleting a network element is shown in Figure 3. After successfully deleting a network element, the network management server sends a message to the performance collection remote service, requesting to stop the performance collection function of the network element. After receiving the request message, the performance collection remote service forwards the message to the performance collection control thread. After the performance acquisition control thread receives the request message, it searches for all created acquisition threads. For each collection thread, it is judged whether the network element requesting to start the performance collection function is included in the collection network element list. If a collection thread includes a network element requesting to stop the performance collection function, the network element is deleted from the collection network element list of the collection thread.

In addition, if there is an extreme situation where a large number of network elements in the system fail and cannot communicate with the network management system, when collecting the performance of each network element, it is necessary to judge the state of the network elements, but this will take a long time. If network elements fail at the same time, the efficiency of performance collection will be seriously affected. In order to solve this problem, a network element status polling thread is started during the startup of the network management server. The specific treatment is as follows:

(i) When the network management server is started, all network elements are obtained from the database.

(j) Acquiring the maximum number of simultaneously collected network elements of each polling thread from the polling configuration file.

(k) Start a cyclically executed status polling process for multiple network elements one by one.

(l) In the status polling thread, the status of network elements is detected periodically, and the polling cycle can be adjusted according to the number of network elements. When the number of network elements is large, the polling cycle can be set longer.

(m) After polling newly discovers a network element successfully, send the request of starting network element performance collection to the performance collection control thread, the performance collection control thread searches for all performance collection threads that this network element type has started, and then judges this Whether the network element already exists in the collection network element list of a certain performance collection thread, if it exists, it will be returned, otherwise, this network element will be added to the number of certain collection network elements that does not reach the configured maximum number of collection network elements per thread In the collection NE list of the performance collection thread. If the number of collection network elements in all performance collection threads currently created has reached the configured maximum value, but the total number of collection threads has not reached the maximum allowable number of performance collection threads, create a new performance collection thread for this network element .

(n) When polling network elements fails, send a request to stop network element performance collection to the performance collection control thread, and the performance collection control thread searches for all collection threads that have been started for this type of network element, and finds the performance of this network element. The collection thread deletes this network element from the collection network element list. If after deleting this network element, the collection network element list does not contain other network elements, stop and delete this collection thread.

Since the polling cycle of the polling thread is generally much shorter than the collection cycle, most of the polling failures can be discovered and notified to the performance collection control thread before the performance collection is performed, thereby improving the collection efficiency.

Considering that network devices often change, when adding a network element, the performance acquisition control thread will automatically assign the newly added network element to a certain performance acquisition thread or create a new performance acquisition thread to control the network element after receiving the corresponding message. When a network element collects historical performance data, when a network element is deleted, the performance collection control thread will automatically traverse each performance collection thread until it finds the performance collection thread containing the deleted network element after receiving the corresponding message. The NE is deleted from the collection NE list of the performance collection thread.

Considering the problem of network communication, the on-off status of network elements may change sometimes. When the network management system starts, it will start the network element status polling thread, which is responsible for monitoring the changes of the network element status. When the status of the network elements changes, the network element status wheel The query thread sends a message of network element state change to the performance collection control thread, and after receiving the message, the performance collection control thread decides to start or stop the performance collection task of the network element according to the state of the network element. If a network element is added, the network element status polling thread will automatically + start the polling task for the network element, and if a network element is deleted, the network element status polling thread will automatically stop the polling task for the network element.

Regularly use the SNMP protocol to access network elements. If no message is received from the network element, it means that the network element and the network management system are disconnected. For unreachable network elements, they need to be deleted from the performance collection thread in time, otherwise the collection efficiency of the performance collection thread will be seriously affected. The method adopted is that when the polling thread finds that the network element has changed from the normal state to the unreachable state, it sends a message to the performance collection remote service, requesting to stop the performance collection function of the network element, as shown in Figure 5. When the polling thread finds that the network element has changed from the disconnected state to the normal state, it sends a message to the performance collection remote service and requests to start the performance collection function of the network element, as shown in Figure 4.

The format description of the configuration file is given below. Considering the extensibility of the configuration file, the configuration file adopts the xml file format. The configuration file mainly includes the following configuration items:

1. Network element type;

2. Whether to collect historical performance of this type of network element;

3. Historical performance collection cycle type 1;

4. Historical performance collection cycle type 2;

5. Collection thread class name;

6. The maximum number of collection threads of this network element type;

7. The maximum number of collection network elements per collection thread of this network element type.

Although the embodiment of the present invention takes the BWA network management system as an example, the method of the present invention is applicable to various network management systems.

Claims (9)

1. A method for realizing historical performance collection in a network management system, comprising the following processes: In the first step, when the network management server is started, the performance acquisition control thread is started; The second step is to obtain various historical performance collection configuration parameters from the configuration file of the performance collection control thread; The third step is to implement a dynamically loadable interface for each performance collection thread; In the fourth step, in the performance acquisition control thread, according to the network element types managed by the network management system, a list of each network element is obtained through multiple database queries; The fifth step is to dynamically create a performance collection thread of this type of network element according to the various historical performance collection configuration parameters obtained from the configuration file of the performance collection control thread for the obtained list of certain network elements; In the sixth step, in the performance collection thread corresponding to a certain network element, each time the timing period arrives, the performance collection is performed cyclically for each network element in the collection network element list of the performance collection thread. 2. the method for realizing historical performance collection in the network management system as claimed in claim 1, is characterized in that: The acquisition of various historical performance acquisition configuration parameters from the performance acquisition control thread includes performance acquisition threads related to network element types, the maximum number of acquisition threads for each type of network element, and the maximum number of acquisition network elements for each type of acquisition thread. 3. the method for realizing historical performance collection in the network management system as claimed in claim 1 or 2, is characterized in that: Before collecting the performance of a certain network element, obtain the state of the network element through the SNMP protocol; If the acquisition of the network element status fails, delete the network element from the collection network element list in the performance collection thread; If after deleting the network element, there are no other network elements in the collection network element list of the performance collection thread, stop the collection thread and delete it. 4. the method for realizing historical performance collection in the network management system as claimed in claim 3, is characterized in that: When the network management system finds a network element, it sends a request to the performance collection control thread to start network element performance collection. The performance collection control thread searches for all performance collection threads that have been started for this type of network element, and then judges whether the network element already exists. In the collection network element list of a certain performance collection thread; Return if it exists, otherwise, add this network element to the collection network element list of a performance collection thread whose number of collection network elements does not reach the configured maximum number of collection network elements per thread; If the number of collection network elements in all performance collection threads currently created has reached the configured maximum value, but the total number of collection threads has not reached the maximum allowable number of performance collection threads, create a new performance collection thread for this network element . 5. the method for realizing historical performance collection in the network management system as claimed in claim 3, is characterized in that: When the network management system deletes a network element, it sends a request to the performance collection control thread to stop the performance collection of the network element. The performance collection control thread searches all the performance collection threads that have been started for this type of network element, and finds the performance collection thread where the network element is located. thread, delete this network element from the collection network element list, if after deleting this network element, the collection network element list of this performance collection thread no longer contains other network elements, stop this collection thread and delete it. 6. the method for realizing historical performance acquisition in the network management system as claimed in claim 1 or 2, is characterized in that, when a large number of network elements break down and cannot communicate with the network management system, the first step further includes: When the network management server is started, all network elements are obtained from the database; Obtain the maximum number of simultaneous network elements collected by each polling thread from the polling configuration file; Start a cyclic status polling thread for multiple network elements one by one. 7. the method for realizing historical performance collection in the network management system as claimed in claim 6, is characterized in that: When the status polling thread finds a network element, it sends a request to the performance collection control thread to start the performance collection of the network element. The performance collection control thread searches all the performance collection threads that have been started for this type of network element, and then judges whether the network element is Already exists in the collection network element list of a certain performance collection thread; Return if it exists, otherwise, add this network element to the collection network element list of a performance collection thread whose number of collection network elements does not reach the configured maximum number of collection network elements per thread; If the number of collection network elements in all performance collection threads currently created has reached the configured maximum value, but the total number of collection threads has not reached the maximum allowable number of performance collection threads, create a new performance collection thread for this network element . 8. the method for realizing historical performance collection in the network management system as claimed in claim 6, is characterized in that: When polling the network element fails, send a request to stop the performance collection of the network element to the performance collection control thread, and the performance collection control thread searches all the collection threads that have been started for this type of network element, and finds the performance collection thread where the network element is located. Delete this network element from the collection network element list. If after deleting this network element, the collection network element list does not contain other network elements, stop and delete this collection thread. 9. The method for realizing historical performance collection in the network management system as claimed in claim 6 is characterized in that: the network management system finds or deletes the network element successfully, and sends a message to the status polling thread to start or stop the status wheel of the corresponding network element inquire.

Images