A problem is described in which a large input data file must be split into many pieces for processing on the Open Science Grid. The chosen solution was to customize the xrootd Posix client library to support a special suffix added to the end of the xrootd url that makes the client library present a shortened version of the input file containing a subset of the data in the original file, book-ended by the mandatory file header and trailer blocks. A test of the modified client under realistic conditions in a grid production environment showed that it was a well-adapted solution. This presentation was made to the XRootD/XCache Working Group on January 7, 2021.
