Simulation tools and experimental data acquisition generate extreme data volumes. Evaluation of these data - which usually means: visualisation of relevant aspects - has bandwidth limits which handicap rapid extraction and processing.
Supercomputer simulate phenomena in space and time with millions of grid points. The result are Gigabytes of data, to be evaluated with visualisation and data reduction tools. However, a number cruncher is too expensive to block it with rendering tasks; data transfer to the client, on the other hand, generates workload on this workstation and floods the network inbetween. Similar problems occur in the analysis of complex experiments: the data rate is too high to allow for a really adequate analysis of the expensive sensor data. Many important evaluations, therefore, cannot be done - or only at high costs.
An example. Let a supercomputer simulation of thermal flow generate 900 GB of data. Shipping these across a 10 Mbit local ethernet takes many hours - impossible to run many varieties of data explorations. Although frequently only a fraction of the volume is relevant, e.g., locations of heat bridges. However, selective data transfer is not supported by todays tools: you always have to send complete files.
Rasdaman is optimized for the rapid extraction from complex spatio-temporal data volumes. Cutouts, feature search, and processing in n-dimensional data spaces are simple to formulate with the SQL-oriented query language. The enabling intelligent database engine accomplishes fastest evaluation. Only those data which are really requested go across the networks to the client.
Gain new insights into your data volumes with rasdaman!