Local, client near data analysis becomes nearly impossible as the data volumes of the CMIP Data Pool exceeds the Petabyte scale. Heavy data transfer volumes would result in expensive and slow workflows which can hardly be sustained due to the growing memory requirements.
Instead, we recommend, promote and support server-side data-near data analysis. The CDP is linked to the High Performance Computer (HPC) file system so that file access is fast and does not depend on external network bandwidth. Furthermore, HPC users benefit largely by the HPC software environment by having access to state-of-the-art version-maintained analysis tools which increase data analysis efficiency.