Crossdata is a distributed framework and a fast and general-purpose data federation powered by Apache Spark. It unifies the interaction with different sources supporting multiple datastore technologies thanks to its generic architecture and a custom SQL-like language using SparkSQL as the core of the project. Supporting multiple architectures imposes two main challenges: how to normalize the access to the datastores, and how to cope with datastore limitations. Crossdata provides connectors which can access to multiple datastores natively, speeding up the queries by avoiding the overhead and the block of resources of the Spark Cluster when possible.