Streaming alarm data analytics for mobile service providers
Engin Zeydan, Utku Yabas, Sozuer Selcuk, Çağrı Özgenç Etemoğlu · 2016
Several thousands of alarm events are arriving into the network operation center of today's mobile service providers per day. On the other hand, managing this huge amount of information is getting much more difficult as the size of the network infrastructure gets larger. The demo presented in this paper proposes a novel scalable architecture for alarm/event data analyses where alarms are captured, processed and visualized with appropriate notifications in real time. Alarm rules are pre-registered offline into the system through root cause analysis module. The proposed system includes novel methods for streaming data analytics with Complex Event Processing (CEP) using platforms such as Apache Kafka and Storm. The presented proof-of-concept demo helps data center and network operators to monitor continuous alarm streams in order to predict the upcoming alarms and potential network failures.