scope:

The present document aims at studying how operations data (KPIs, metrics, alarm notifications, event logs, debug information) can be exploited to ensure the availability and reliability of NFV-MANO and of the network services it manages using data analysis/data driven techniques. This includes, among others, the use of machine learning to find patterns for cases where detailed semantics information is unavailable (e.g. due to confidentiality) or the amount of data is overwhelming. Use cases are created describing how the information can be used offline (for example for root cause analysis and predictive maintenance resulting in, e.g. creation and/or changes of deployment flavours) and online (to identify appropriate LCM operations and policy changes in order to achieve the intended service availability objectives).