Industrial and scientific datasets have been growing enormously in size and complexity in recent years. The largest transactional databases and data warehouses can no longer be hosted cost-effectively in off-the-shelf commercial database management systems. There are other forums for discussing databases and data warehouses, but they typically deal with problems occurring at smaller scales and do not always focus on practical solutions or influencing DBMS vendors. Given the relatively small (but highly influential and growing) number of users with these databases and the relatively small number of opportunities to exchange practical information related to DBMSes at extremely large scale, a workshop on extremely large databases was organized. This paper is the final report of the discussions and activities at the workshop.