Securified

We address efficiently the needs of confidential-preserving analytics on large data sets in third-party clouds. Based on over 10+ years of research experience in confidentiality-preserving data processing and over various funding, our solution is centered on a novel notion of extended security policy (patent application in process). Based on Spark, we allow not only different custom confidentiality levels to be defined for data with differing secrecy requirements, but also to assign security mechanisms to these. Coupled with a core mathematical model of (the behavior of) queries executed by our system which allows for tracking data flows, this allows reasoning about security constraints at fine granularity of single operations and thus automated selection of most suitable mechanisms. This enables vastly increased performance (10x over existing systems) while also providing formally verified precise guarantees (Securified = Security + Verified) as well as interoperability compared to competitors.