While Peer-To-Peer (P2P) networks have evolved largely outside the academic community, they became increasingly popular with computer science researchers lately due to their compelling properties like robustness, scalability and resistance to failure. Due to lack of centralized control a major problem in any P2P system is the location of suitable information sources, a problem especially challenging in dynamic scenarios, where nodes join and leave the network regularly. In order to meet these challenges we developed an architecture for query processing which we call “The Query Web”, based on a hybrid P2P index combining a distributed hash table approach with an unstructured P2P network. By assigning the high-level indexing task to the reliable nodes in the P2P network, we expect the index to be reasonably insensitive to both changes of the network structure as well as updates of the data instances.