The Construction Of Data Center Operations Data Association Rule Knowledge Base

With the rapid development of information technology, expanding the size of the cluster, the system operation and maintenance technology put forward higher requirements. In order to protect the operation of the system and the system health monitoring, operation and maintenance personnel to run the information collection systems across multiple dimensions, operation and maintenance of these data to maintain the system. However, operation and maintenance personnel are usually the operation and maintenance of the various dimensions of the data were analyzed separately, there is no analysis of the potential relationship between the various dimensions of operation and maintenance data. Each dimension accumulated vast amounts of data operation and maintenance, lack of effective data analysis and mining tools, not the collected data for effective operation and maintenance of analysis used.In order to tap the potential value of the operation and maintenance of data, operation and maintenance to improve the means of operation and maintenance personnel. This article describes the association rules based on the operation and maintenance of data mining and knowledge base to build technology. Capable of operation and maintenance of multi-source data mining and correlation analysis depth to build a knowledge base describes the correlation relationship and the causal structure of multi-source data between the operation and maintenance, so that operation and maintenance personnel to better understand the internal operation of the system, and means to enhance the efficiency of operation and maintenance. In the process of mining association rules to solve the two difficulties:category tags and association rule mining system log improve the algorithm.When the data mining association rules, required by category of data mining association rules, and the system does not log category attribute, so it needs to be labeled category. System Log category tags clustering and classification into two processes. Clustering process is complete log category feature extraction, feature classes build knowledge work. The classification process is not carried out to complete the log category tag, category by category to match the characteristics of the knowledge base mark work. This technical knowledge can be characterized according to the category of logs for accurate classification, classification accuracy is much higher than traditional text clustering and classification methods. By detailed theoretical analysis and experimental results on the large-scale log data to prove the effectiveness of the invention.Proposed a multi-source operation and maintenance of inter-association rules mining method, the operation and maintenance of multi-source data into the system event data, the timing sequence of system event data consisting of association rules mining. To make more use value association rules, association rules during mining, the need to retain timing information between related items. This article describes the improvement on Apriori algorithm, mainly in the way of counting rules Apriori algorithm support and generate k+1 candidate sets were perfect, thus ensuring the timing information between the various associations association rules to meet the operation and maintenance of multi-source data analysis requirements associated with the scene.