As an optimization, Kobe keeps copies of populated Virtuoso and other databases so that they do not need to be populated from scratch for each experiment. This is currently done on disk space served by NFS, although the files are not directly accessed on NFS but downloaded to the containers' local space and accessed there. It would be more efficient to use an FTP server or similar for this.