Big Datacube Analytics with rasdaman: Flexible, Scalable, Secure

P. Baumann
Rasdaman GmbH, Bremen, Germany

Keywords: Big Data, datacubes, analytics, common information space, secured access

Datacubes are an emerging paradigm for homogenizing and aggregating massive spatio-temporal data into analysis-ready objects. As spatio-temporal sensor, image, simulation, and statistics data datacubes appear in all science and engineering domains, and beyond. Examples include 1D sensor data, 2D satellite imagery, 3D x/y/t image timeseries and x/y/z geophysical voxel data, and 4D x/y/z/t weather data. The rasdaman ("raster data manager") datacube engine offers interactive analytics through server-side evaluation of high-level queries enabling users to ask "any query, any time, on any size" based on open standards interfaces. At the same time, rasdaman is the only datacube tool enforcing easy-to-define spatio-temporal access control down to pixel level. Scalability boosters include CPU and GPU parallelization, federated clouds, and planetary-scale datacube fusion. Single queries have been split across 1,000+ cloud nodes. Benchmarks of the 2018 RDA Array Database report show that rasdaman can be 304x faster than other technology. Altogether, this software "made in Germany" is multi-Petabyte practice-proven, multi-award winning (ex: 2018 NATO Defence Innovation Award), and worldwide recognized technology leader.