The range delete is surely faster, but you'll be better to check out
the actually deleted keys from the returned array, or assume some keys
may have not been deleted (.i.e. because part of the ensemble was
unavailable or the key didn't actually exist).
On the other side, with a range get followed by a one-by-one delete,
you could immediately know each delete response.
That's the tradeoff.
Hope that helps,
Cheers,
Sergio B.
--
Sergio Bossa
http://www.linkedin.com/in/sergiob