We have built an infrastructure for the {\em lecture on demand} application.This infrastructure consists of a video capture card,the Virage VideoLogger system, our Java-based indexing framework called jetMAX, and the Virage Video Application Server. For a recorded lecture, we first digitize the video signal using the capture card and transform it into an MPEG-2 stream. The MPEG-2 files are then read and analyzed by the VideoLogger system. In addition to the included audio analysis, we implemented two new analysis plug-ins for color and texture feature extraction. The extraction of all features happens in real-time. Therefore, the analysis of a lecture video can be completed within 1-2 hours and results in about 5MB of raw vector and text data. This data is then stored in our indexing framework jetMAX which supports a variety of indexing schemes, such as R-trees and X-trees. In order to process user queries, we facilitate Virage's Video Application Server that provides web-based multimedia search and retrieval interfaces. For a given user query, we issue the corresponding subqueries to the index structures for color, texture, etc. and combine the results. The answer is then presented to the user as a list of video shots ranked by relevance from which she can pick the ones to be played back. The playback of the videos is also handled by the Video Application Server.