hi Eric,

If you are on OPS, you can just specify the DSMO_NODE in EBU_ADDRLIST to
use different node name.  If you are not on OPS, the only way I see to use
your approach is to keep track of which tablespace/datafiles are backed up
with which node name like you mentioned.  If you are using this approach,
you will need to keep good accounting of what you backup with which node
name.  If you can keep good accounting, I think the performance will be
better.  As for archive logs, I think, you should use the same node name to
backup.

regards,
Thiha

>In my environment, data is backup from client to ADSM disk pool then
>migrate to tape with collocation. In consideration of client restore, to
>improve performance, it will be better to have data reading from more
tapes
>in parallel, Thus, to prevent every backup files go into one node thus one
>tape, I am thinking of using more than one nodename for the EBU backup.

>Question: Is it the right approach, if so how to do the coding? Should I
>split the EBU backup script into several each take care a number of
>datafiles/tablespaces; and for each script export a different DSMO_NODE
and
>run them concurrently for backup & restore.  Thank you.

>Regards,
>Eric Tang