Preparation / Naming

When archiving the data (on tape) one should account for common boundaries involved in the archiving process. For the dCache tape pool, file operations on plenties of files of small size (up to some megabyte) are very time consuming and are significantly slowing down data transfers. Very large files of size of some hundred gigabyte will most likely split on several tapes, also slowing down the process due to tape changes involved. Hence, for the dCache tape storage the creation of an archive file of decent size is most suitable with respect to time consumption. Beside time aspects, the creation of meta-data for an archive can become complicated - if, for example, data of different specimen will be combined in a single archive file.

Thus, the data migration tools available operate in a specific manner with respect to data directory structure and directory size, see sec. 1.4. Whenever possible one should provide the data foreseen for archiving in an appropriate directory structure. For beamlines at PETRA III an appropriate directory tree may look as shown below. Depending on the overall amount of data, it will result in a single archive file or several archive files corresponding to a single specimen or scan.

.1 /PETRAIII/pxx/2013A. .2 name_for_a_beamtime (e.g. 1000XXXX_name). .3 online. .4 image_00001.fio. .3 sample1_or_scan1. .4 pilatus. .5 image_00001_00001.cbf. .5 image_00001_00002.cbf. .4 pco. .5 <empty>. .4 lcx. .4 perkinelmer. .3 sample2_or_scan2. .4 pilatus. .4 pco. .4 lcx. .4 perkinelmer. .2 name_for_another_beamtime. .3 online. .3 sample1_scan1. .3 sample1_scan2. .3 scan3. .3 scan4.