Server-side visualisation of massive datasets

Server-side visualisation systems comprise of a thin client connected to a rendering service located on the data server. We present a systematic and quantitative method for evaluating server-side visualisation systems. We evaluate the fundamental limitations of such systems in relation to network performance. Using these results we benchmark four common architectures by implementing clients and/or servers in each system. The suitability of each architecture is assessed for several applications, in terms of scalability, performance and ease of development