On 14/06/13 20:26, Tim Mooney wrote:
> 
> Recovers, of course, are another matter -- there's really no way to dedupe
> a recover, so all the data must come back somehow.  That's one thing that
> I also asked about in the thread I started in November 2012.

Speeding up recoveries requires innovative thinking. In the virtual
world you could have a system where a disk based backup target (say for
example a Data Domain) could host the recovered image on the same media
as the backup in order to have it up and running in seconds before
migrating the data back to a more usual storage location.

Now wouldn't that be clever?  ;-)