With the development of the information communication network equipment functions and business services

the role of the monitoring department has become an increasing important in the daily work of network maintenance. In this paper

use the k-means algorithm of data mining to make cluster analysis with the original alarms which come from six months

discussing the universal rule of alarm restoration

establishing mathematical mode with discrete and continuous ways to study the setting of the optimum time

offering some ideas and suggestions for the alarm management and the use of big data technology.