There are many causes of network faults which are quite difficult to detect by syslog or trap information of servers/network devices.
Operators can not notice an impact of a silent fault even when the users of the service are aware of the trouble.
Analysis from various sources like traffic data, natural language data of SNS, etc.. could make it possible.
In this program, we'd like to introduce analysis examples in which detection of a sign of a fault and estimation of the extent of its impact are enabled with referring actual cases of contents providers.