In this paper we are presenting a novel approach that enables rendering large-shared datasets at interactive rates on a remote inexpensive workstations. Our algorithm is based on view-dependent rendering and client-server technologies. In our approach, servers host large datasets and manage the selection of the various levels of detail, while clients receive blocks of update operations which are used to generate the appropriate level of detail in an incremental fashion. We assume that servers are capable machines in term of storage capacity and computational power while clients are inexpensive workstation that have limited 3D rendering capabilities. To avoid network latency we have introduced two powerful mechanisms that cache the adapt operation blocks on the clients’ side and predict the future view-parameters of clients based on their recent behavior history. Our approach dramatically reduces the amount of memory needed by each client and the entire computing system since the dataset is stored only once in the local memory of the server. In addition, it decreases the load on the network as a result of the incremental update contributed by view-dependent rendering.