Study On Network Management By Classification Of Datamining

The goal of network management is maximum increases the network available time, enhances the network equipment operating factor, the network performance, the grade of service and the security, provides the network the long-term plan.The attack and the viral influence which along with the network scale unceasing expansion, user's fast growth, emerges one after another incessantly, the network management and the maintenance becomes more and more important, more and more difficult.The traditional network management software enables the network superintendent in the dynamic observation network equipment condition and the network position, but is unable to obtain breakdown connection between each equipment, is unable to obtain during each agreement performance influence, similarly is unable to discover the hideaway"softness"the breakdown.These"soft"the breakdown often creates by the network attack, and chronic spread, and often in has created before the substantive network paralysis, already ambush very long time.How can the breakdown produce the initial period after"softness"to forecast the substantive network breakdown occurrence, this to manages the network to have the vital practical significance.The network management system, IDS, Firewall have the massive network performance data and the warning every day.In these data has massively latent about the network movement process in, each equipment performance data, these data are the network performance real reflections, is concealing the real situation which between the network various elements relates.The knowledge obtains which based on these data has the vital significance to the network performance analysis and the maintenance.Therefore we must discover the new method research network performance, but the data mining method to the network performance knowledge discovered has the extremely vital significance.How"afterwards"will the processing strategy become"beforehand", how will be able to forecast the question the appearance, then will be this topic research goal.Synthesizes, the present paper has mainly carried on following several aspect work.In view of the topic, the paper mainly stresses on through to the network data collection and the analysis, designated the suitable data mining algorithm, and will classify the algorithm concrete application will be the decision tree algorithm, the key point is solves the excavation method choice and the multi-data pool data warehouse production.Through to the network data analysis, discovered must achieve the forecast network breakdown the goal, gathers the traditional network management data is merely insufficient, needs to unify IDS and firewall other in view of the network attack warning data.After the union data pool, after satisfies the union the data pool, has met the system need.We have introduced improvement decision tree algorithm C5, carries on the classification with this algorithm as the data mining method to the breakdown information, achieves the breakdown diagnosis and the forecast goal.In this foundation, has studied the network management system architecture and the diagnosis method, had determined the pattern classification uses each kind of performance data information. In this foundation, studied the system to need the data pool, had determined used each kind of data message.In the article the 1st chapter proposed in the cash network management exists question as well as based on data mining technology network management system; 2nd chapter through the analysis data mining and the network management technology theory, proposed the different discovery pattern to the network breakdown diagnosis forecast compatible question, and gives the answer, this for article in a key point; Through the front analysis, the 3rd chapter discovered the method decides as the classified method the decision tree algorithm, and the basis network data characteristic, has given about the network storm example decision tree production process; In the article another key point is the data pool choice and the data warehouse production question, this in 4th has made the detailed elaboration with 5.3rd; 5th chapter mainly is to system realization frame statement, with each main question explanation; 6th chapter has built the experimental environment and has obtained the anticipated experimental result.