An outlier in time series data is often a signal that must be addressed. Domains where outliers detection can give noteworthy informations are various: -Technical supervision -Cybersecurity, fraud detection -KPI business -. . . At Orange, we developed Astrolog to detect outliers from our time series data and analyze unexpected behaviors. Linked to Astrolog, Astropolis can trigger different levels of reactions according to the situation, such as log anomalies, send alerts by mail, stop some processes, .. During the past months, Astrolog and Astropolis helped our users to detect early some discrepancies and trigger very quickly the right reaction avoiding potential dramatic consequences Spark and Cassandra are capable of handling the challenges that might arise for this use case : massive scalability, high availability and high performance. In this session, I will show how we are using Cassandra and Spark for analyzing time series data and consequently trigger the right actions About the Speaker Jean Armel Luce Tech Lead, Orange Jean Armel is a Database Tech Lead at Orange, with more than 20 years of software development in various environments. For the last 5 years, he has been using Cassandra for many applications that require scalability and high availability. Jean Armel has worked with Apache Cassandra since 2011 and is a regular speaker at technical conferences in France, London or San Francisco.