Mapping user-level QoS to system-level QoS and resources in a distributed lecture-on-demand system

Information repositories and database systems that combine time-independent and time-dependent data will be an integral element of the future information infrastructure. A seamless integration-especially with respect to QoS-of (multimedia) database systems is of central importance. Furthermore, response times are very important for these systems, but the particular QoS to be supported and the amount of needed resources depend on the query result. Thus, QoS negotiation and resource reservation can only be performed after the query result is known. In our solution, we extract QoS related information from the meta-data and use this knowledge for QoS guided query processing, QoS negotiation between multimedia database system and multimedia object request broker, and for early resource reservation. This approach enables us to select an optimal middleware configuration and to allocate the appropriate amount of resources in all system components without increasing initial response times.