While most available client-server advanced visualization products offer good interaction performance in a local LAN environment, it is very challenging to maintain productive user interaction when the communication occurs over the Internet. This is especially true when long round-trip latencies (over 200msec) are involved, such as when using a cellular modem or over large physical distances. Many ThinsightTM design features, some of them unique in the industry, help it overcome these challenges:
The user interface logic runs entirely on the client, eliminating all round-trips to the server other than the ones actually requiring large amounts of computations.
Large data volume loading and results saving occurs only inside the server. Efficiently compressed images are transmitted across the client-server connection.
Data transmission, rendering and display tasks are pipelined, allowing multiple computation cycles to overlap, typically providing 2-3 times faster screen update rates than competitive non-pipelined architectures.
Internal self-measurement mechanisms continuously monitor the operational environment of the system, including upload and download transfer rates, transmission delays, and computation time of key tasks. These parameters are then used to control various quality/speed tradeoffs along the computation pipeline, such as the rendered image resolution and degree of compression being used during interactions (the final image quality is under user control).
By adding high-contrast text and graphics only at the client, rendered images can be sent from the server in a highly compressed format without introducing strong compression artifacts.
