HEAD requests of huge files will be a bit more inefficient.
See discussion at https://unix.stackexchange.com/q/331611