Allow ADAORD to work two blocks at a time to slowly crawl through a file. only the records from the first two RABNS would be put on hold and movement of records from #2 to #1 would be coordinated with the nucleus. Once the first block was finished, all the records would come off hold and then the process could repeat with block #2 as the receiving block an #3 as the giving block.
It could provide a staring Rand ending ABN number or have a limit of blocks processed. this would allow for partial reorders while not interfering with current operations. it could even allow for multiple runs against various RABN ranges of the file at the same time. This could get a large file reordered faster then a single front to back approach.