The complexity and sophistication of large scale analytics in science and industry have advanced dramatically in recent years. Analysts are struggling to use complex techniques such as time series analysis and classification algorithms because their familiar, powerful tools are not scalable and cannot effectively use scalable database systems. The 2nd Extremely Large Databases (XLDB) workshop was organized to understand these issues, examine their implications, and brainstorm possible solutions. The design of a new open source science database, SciDB that emerged from the first workshop in this series was also debated. This paper is the final report of the discussions and activities at this workshop.