Due to the proliferation of tools and techniques for creating, copying and sharing digital multimedia content, retrieval by content is becoming more and more important, for example for copyright protection. Recent advanced multimedia descriptions typically describe the media through many local descriptors, which both increase the size of the descriptor collection and require many nearest neighbour queries.  Needless to say, scalability of query processing is a significant concern in this new environment. The typical “computer-vision-based” approach to dealing with scalability is to compress the descriptors and fit them into memory. We argue, however, that even with compression the data will eventually overflow memory and therefore investigate a "database-based" solution to the scalability issue. This talk will give an overview of the major scalability results of our project.