>Dwight Cook wrote:

>    Well actually if you would do something like...
>
>    nohup restore /filesys1/* -subdir=yes -replace=yes -pass=blah &
>     nohup restore /filesys2/* -subdir=yes -replace=yes -pass=blah &
>     nohup restore /filesys3/* -subdir=yes -replace=yes -pass=blah &
>     ...
>     nohup restore /filesysn/* -subdir=yes -replace=yes -pass=blah &
>
>    you could reduce your restore time IF the data doesn't all reside on
>     the same tape (also depends on your network, client processor, yada,
>     yada, yada...)
>
>     Dwight

We also have the need to restore multiple filespaces and noted the example
in the IBM documentation stating that multiple commands such as
       dsmc restore /u/jones/ ...
       dsmc restore /u/smith/ ...
       dsmc restore /u/brown/ ...
is faster than one command such as
       dsmc restore /u/ ...

We assume there is a gain in cputime from multiple processes.
We probably can understand it may not be that great if the data all
resides on the same tape (as noted by Dwight), but would it be just
as bad if one has to keep mounting/unmounting tapes (we have only onedrive). So 
is there any real advantage with multiple restore commands as advised in the
IBM documentation? Any breakdown on the time saved will be greatly
appreciated.

francis yeung
Daresbury Laboratory
CCLRC