Design And Implementation Of Rule-based Management Of Data Collection Task

Network equipments (such as hosts, switches, routers, etc.) are likely to maintain some data which are related to their own status or network, and provide some command interfaces which use some sort of protocol to collect data. Network managers can collect and view the detail data of devices on all aspects through those command interfaces. After that, network managers can know the running status of equipment, information of network status, and can manage network equipments.However, a wide range of network equipments and the variety of manufacturers, types, models, operating systems of equipments may affect the collection method on different equipments, and the ways to analyze the report of the collected data and so on. Moreover, different manufacturers have different specific management interfaces and a variety of private information. Traditional data collection system was developed for the equipments of specific type, and fixed all private information in the program, so that when a new type of equipment was produced, it was necessary to develop a new data collection system.Because of this, not only the collection methods for different equipment are different, but also the data collection task for the management process is different because of difference of equipments. What's more, even for the same equipment, because of using different protocol to collect, the data collection task for the management process is also different. This requires not only the general of collection process of data collection system, but also the general of data collection task. The execution of data collection is task-based. If the management to the tasks of data collection is also achieving general, the initialization, the dispatch and the display of data collection task's process would have a higher efficiency and wider use of space.However, because of data collection system achieving general, the process of task management for data collection, which includes the initialization, the dispatch and the display of a data collection task, must have a series of logical estimation to the used protocol, the type of equipment and so on. With the expansion of the network, the functions of the systems are more and more complex. At that time, those logical estimations may also be more difficult to manage and upgrade.This paper is under the background of the project, which is named"Data Collection System". In the data collection task management subsystem, the integrating of rules engine DROOLS not only made task management subsystem achieve general, but also made full use of the advantages of rule engine DROOLS to realize business logic and process's separation, which can better manage the logic. The logic may expand increasingly and be more complex. At the same time, the system also has greater flexibility and higher efficiency, which provides a broad space for the upgrading of task management subsystem of general data collection system in the future.Task management subsystem of"Data Collection System"is divided into task initialization module, task dispatching module and task display module. There are a series of logical estimation in these three modules, so it uses a certain degree of DROOLS rules engine to solve those logical estimations.Data collection task initialization is the first part of data collection task management subsystem. The process of execution of data collection task in the Data Collection System also starts from data collection tasks initialization. A complete collection task requires three messages as follows: the name and the path of task template, specific parameters of equipment and the dispatching strategy. What is more, the launch of the collection task is the last step, so there is a total of four steps in process of completing a collection task initialization.The task template contains some information including the methods of data collection, data partitioning, data processing, and data forwarding of some types of network equipments. It also includes the way of collection (one-time collection or periodic collection). When customizing data rule and logic rule, the choice of data's fields is also pointed out the types of data, which we want, so the whole process of data collection, in a manner, is based on the content of the task template. A task template file is also a rules file, which was pre-customized in the client. In order to allow users to understand the information of the selected task template, the system needs to read and analyze the rules files of task template, and displays that information on the interface, which can guide the users to complete choosing task template, in the task initialization process.The process of starting the collection task through the client interface is a process which has packed the information of task configuration into an UDP packet and sends the packets to the dispatch centers. The method of sending task configuration information to the dispatch center is different in different tasks, as well as the collector and the equipment needed for different types of parameters. So the system will classify the tasks according to the contents of the tasks, and then pack the task configuration into UDP packets in different ways. This stage contains a certain degree of logical estimations, so DROOLS rules engine is integrated and used to match the contents of the collection tasks and judge whether it is matched, which can guide the work of making UDP packets containing the task's parameters and so on. At recently, this system differentiates the way, relatively simple, to send the packets containing configuration of tasks according to protocol. So each protocol has a rule which guides to make and send the packets.In the whole process of task initialization, program maintains an object instance that contains the combination of task's parameter. When it is put into the working memory of the client, the corresponding rules are activated. While the rules are executing, they get the specific parameters of task under current protocol from the object instance, and put these specific parameters into the UDP packets. This object instance of task's parameters will be a basis for rules matching and judging in the task display module. Even the UDP packets have been sent, the object instance is still preserved in client's working memory.Data collection task dispatch is the second part of data collection task management subsystem. Task dispatch is a process, in which dispatch center send the packets to a suitable collector according to the dispatch strategy. The packets include the parameters of data collection task initialization in client. In this system, there are four dispatch strategies which can be supported. They are"the designated collector","random collector","the smallest number of task collector", and"the smallest weighted average collector". As for what the current task dispatch strategy is, is decided by the process of the task initialization which is set by user.In the process of task dispatch, the different dispatch strategies have different ways of selecting collector and different processes of dispatch, so there are a certain degree of logical estimations in this stage. This work is also completed by the rules engine DROOLS. At recently, the dispatch strategy being contained in the task parameters packets has become the matching basis of the task dispatch's rules. At first, task dispatch's rules judge the type of packets from client or collector to see whether the packets of arrival included in the task initialization parameters, and then match the type of dispatch strategy contained in packet. There are four types of tasks dispatch rules corresponding to the four dispatch strategies used in the task dispatch module of the system now. Finally, the task dispatch strategy and the collection task's parameters are transferred to the dispatch center. The work to choose the collector and tasks of distribution is not completed in the rules, but in the dispatch center.Data collection task display is the third part of data collection task management subsystem. The function of task display is to display the real-time information, status and changes of task through the client's interface. It is also called the updating of task information. At the beginning the process of task display, client analyzes packet which is sent from the dispatch center, and then match and judge the contents of the packet to see whether it includes task display command. If so, what is the type of command (create a task, add task, delete task, suspend task or change task information, etc.). Because of the difference of task display commands, the process of task display and interface updating are different. There are a certain degree of logical estimations on contents of packets in this stage, and rules engine DROOLS is also integrated. As said earlier, the task display module and tasks initialization module maintain the same rules working memory together, since for completing the task initialization, client needs feedback that whether the tasks is successfully launch. If the object instance of task parameters'information in the working memory matches the task display commands in the packets from dispatch center, and if the packet contains the information that the task is running, the task is successfully initialized. Only then will the system be able to delete the object instance of task parameters'information from the working memory. The task display rules may match and judge the information, and activate the corresponding rules. After the rule has been activated, the corresponding update method will be called and a variety of operations of the list of tasks in the client interface would be carried out. So as the real-time information of tasks would be presented in the interface.