Amusingly, the reason that NIH uses FTP, rather than HTTP, for bulk genome file transport is that because in the old, old days, one PI (professor with a grant) was running an old version of IE which couldn't download files larger than 2GB. They had a lot of trouble with file corruption and because the PI was "important" NIH hasn't really moved to an HTTP content distribution method. Mirroring the NIH FTP servers was always a huge pain.